Chemical Engineering Science 59 (2004) 623 – 632 www.elsevier.com/locate/ces Time-dependent gas holdup variation in an air–water bubble column Chengzhi Tang, Theodore J. Heindel∗ Department of Mechanical Engineering, 2025 Black Engineering Building, Iowa State University, Ames, IA 50011-2161, USA Received 7 August 2003; received in revised form 17 October 2003; accepted 22 October 2003 Abstract Time-dependent gas holdup variation in a two-phase bubble column is reported with air and tap water as the working uids. The results indicate that time-dependent gas holdup is closely related to the water, whose quality is unsteady and changes, not only during the two-phase ow, but also during idle periods. The signi cance and characteristics of the time-dependent gas holdup variation are in uenced by the bubble column operation mode (cocurrent or semi-batch), the sparger orientation, the super cial gas velocity, and the super cial liquid velocity. It is proposed that a volatile substance (VS), which exists in the water in very small concentrations and inhibits bubble coalescence, evaporates during column operation and results in a time-dependent gas holdup. The in uence of bubble column operation mode, sparger orientation, super cial gas velocity, and super cial liquid velocity on the time-dependent gas holdup variation are explained based on their e ects on bubble size, bubble contacting frequency and mixing intensity. This work reveals that regular tap water may cause signi cant reproducibility problems in experimental studies of air–water two-phase ows. ? 2003 Elsevier Ltd. All rights reserved. Keywords: Bubble column; Coalescence; Hydrodynamics; Multiphase ow; Transient response; Voidage 1. Introduction Gas–liquid two-phase bubble columns are the subject of much research, and water and air are used in a majority of these studies. It is usually assumed that the properties of air and water are steady and the results obtained in these air–water systems are reproducible. However, Anderson and Quinn (1970) compared gas holdup in a 21 cm ID semi-batch bubble column using distilled water, deionized water, and tap water; they found that at the same super cial gas velocity, the gas holdup in tap water was up to 50% higher than in distilled water, while the gas holdup in deionized water was intermediate to the former values. They also found that solutions made from mixing varying quantities of tap and distilled water also gave di erent gas holdup results at the same super cial gas velocities. In addition, they observed that heating tap water could change the gas holdup and regime transition in the bubble column. They inferred that there were some bubble coalescence-inhibiting impurities in the water and the concentrations of these impurities di ered among water types. They concluded the gas holdup ∗ Corresponding author. Tel.: +1-515-294-0057; fax: +1-515-294-3261. E-mail address: theindel@iastate.edu (T.J. Heindel). 0009-2509/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2003.10.016 variations were the result of the coalescence-inhibiting impurities and ordinary purity precautions did not necessarily result in reproducible results. Using tap water and air from a compressor, Maruyama et al. (1981) noticed that di erent gas holdups were obtained in each of the 3 experiments repeated in the same 2-D semi-batch bubble column without ltering the air or changing the water. They attributed the di erence to the accumulation of trace impurities, such as compressor oil, in the water. In other experiments reported in the same paper, they found that impurities of di erent concentrations could change the ow transition in bubble columns. Ueyama et al. (1989) revealed that types of water (tap water or ion-exchange water) and gas (compressed air or N2 ) could change the gas holdup and ow behavior in a semi-batch bubble column. They hypothesized that there were some coalescence-inhibiting impurities in the water and the concentrations were higher in tap water than in ion-exchange water. They also inferred that the presence of a mist in the air, believed to be oil droplets, hindered bubble coalescence as the mist accumulated in the water. They believed that the coalescence-hindering impurities in the tap water, together with the coalescence-weakening mist in the compressed air, caused a rather complex hysteresis behavior observed in the bubble column. 624 C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 Tap water is a complex mixture consisting of many components. Most of the components exist in very small concentrations, some of which are too small to detect with common measurement devices. Even in ion-exchange high-quality water, there are contaminants due to the ion-exchange process and facility (Michalson, 1968). Surfactants may also leach into water during storage and=or transportation. Skjevrak et al. (2003) reported that volatile organic compounds could migrate into drinking water from plastic pipes. There is a large amount of literature showing that certain organic compounds or surfactants in very small concentrations in water could dramatically inhibit bubble coalescence (Jamialahmadi and Muller-Steinhagen, 1992; Krishna et al., 2000; Zahradnik et al., 1999; Camarasa et al., 1999; Ueyama et al., 1993). Gas holdup in a bubble column can also be in uenced by the gas distributor, super cial liquid velocity, and super cial gas velocity. Park et al. (1977), Miyahara et al. (1983), Terasaka et al. (1999), and Snabre and Magnifotcham (1998) studied bubble formation at gas distributors as well as the e ects of distributor design and super cial liquid velocity on bubble size. Camarasa et al. (1999) studied the combined e ects of gas distributors and coalescence-inhibiting surfactants on gas holdup. Kelkar et al. (1983) pointed out that in a cocurrent bubble column, the gas distributor plate could have a signi cant e ect on gas holdup at low gas velocities. In this work, the time-dependent gas holdup change in an air–water bubble column is presented. The bubble column is operated in either cocurrent or semi-batch mode. The time-dependent gas holdup variation is primarily attributed to the presence of a coalescence-inhibiting volatile substance (VS) in the water. The experimental results show the in uence of column operating mode (cocurrent or semi-batch), sparger orientation (ori ces face upward or downward), and gas and liquid ow rate on the time-dependent gas holdup behavior. 2. Experimental procedures The experiments for this study are conducted in a cylindrical cocurrent bubble column, which consists of four 0:94 m tall acrylic tubes with 15:24 cm internal diameter. Five delrin collars, each 5:1 cm tall, and 11 buna-n gaskets are used to connect the acrylic tubes for a total column height of H = 4 m. Fig. 1 shows a schematic of the entire system. Air is pumped by an air compressor through a lter and enters the bubble column from the bottom via a spider sparger. The air owrate is adjusted with a regulator and measured with one of three gas owmeters, each covering a di erent owrate range. Water from a 379 l reservoir (ID: 0:882 m; material: linear polyethylene) is pumped into the column. The pump is connected to the reservoir with a 2:44 m long 7:62 cm diameter PVC pipe. A 2:85 m long 2:54 cm diameter PVC pipe connects the pump to the column. The liquid owrate is measured with a magnetic owmeter and varied via a pump power frequency controller. Water enters the column through a ow expander and ow straightener, both made of acrylic, to obtain a uniform liquid velocity eld at the entrance region prior to the spider sparger. A gas–liquid separator is located on top of the column where air is separated from the water while the water returns to the reservoir through a PVC pipe. The column is operated in semi-batch mode when the valve at the column base is closed after a given height of water lls the column. Along the column, 5 pressure transducers (P1–P5) are installed, one in each of the ve delrin collars. Each acrylic tube section is numbered 1– 4 from the bottom of the column. Two type-T thermocouples are also located at the bottom and top of the column, respectively. All pressure, owmeter, and thermocouple signals are collected via a computer controlled data acquisition system. Supercial gas velocity and super cial liquid velocity are controlled by the gas regulator and power frequency controller, respectively. The spider sparger, shown in Fig. 2, has eight arms made of 12:7 mm diameter stainless-steel tubes. Thirty-three 1:6 mm diameter holes are located on one side of each arm and distributed as shown in Fig. 2. The arms are soldered to the center cylinder of the sparger such that all the ori ces face to the same direction. Air enters the spider sparger from the central cylinder and exits from the arm holes. The sparger can be installed in one of two ways, with the ori ce side facing either upward or downward. All experiments in this study are carried out under atmospheric pressure and ambient temperature. Tap water and air from a compressor are used as the working uids. Two procedures are used to identify the time-dependent gas holdup variation. In Procedure I, the super cial liquid velocity is xed and the super cial gas velocity is increased gradually from 0 to 20 cm=s. The super cial liquid velocity is then changed and the super cial gas velocity is varied again from 0 to 20 cm=s. A gas holdup versus super cial gas velocity curve is obtained for each super cial liquid velocity. This procedure was repeated over several days using the same water. The di erence between the curves, if any, reveals a day-to-day gas holdup variation. The process is repeated until the curves obtained in two consecutive days overlap, after which “steady state” is de ned. In this procedure, the interval between each measurement is 5 min, not including the sampling time; for each data point 4800 readings are collected every 10 ms and averaged to get the recorded values. In Procedure II, both super cial liquid velocity and super cial gas velocity are xed while data are taken every 2 min over a long time period. In this procedure, 1200 readings are collected and averaged for each data point with the sample time set as 10 ms. A gas holdup versus elapsed time curve is then determined for the speci ed condition. The time-dependent variation in gas holdup is directly C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 625 Fig. 1. Schematic of the cocurrent bubble column experimental facility. Fig. 2. Schematic of the spider sparger. observed in this procedure. “Steady state” is de ned when no signi cant increase or decrease in gas holdup with time is observed. A remark should be given on “ xing” experimental conditions: in Procedure I, xing super cial liquid velocity implies slight adjustments to the pump frequency 626 C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 Fig. 3. Time-dependent gas holdup variation in the cocurrent bubble column obtained via Procedure I. controller to maintain a nominal super cial liquid velocity whenever the gas owrate is changed. In Procedure II, xing simply means setting the pump frequency controller and gas regulator to a speci ed condition to set the super cial liquid and gas velocities, respectively. These settings remain unchanged although the actual magnitudes of both super cial velocities change slightly due to the gas holdup variation in the bubble column. However, the small changes in gas and liquid owrates are believed to be the results of, rather than the reason for, the gas holdup variation. Each time an experiment is initiated, attention is paid to prevent water from entering the sparger so that the bubble generation process would not be in uenced by water inside the sparger body. This is implemented by turning on the gas ow before water rises to the sparger. With ve pressure signals, the time-average gas holdup in each section is calculated from i = 1 −
pi ;
p0; i (1) where
pi = pl − ph is the pressure di erence between the lower (pl ) and higher (ph ) ends of column section i (i = 1; 2; 3; 4), and
p0; i is the corresponding static pressure di erence when the column is lled with water only. Eq. (1) neglects the e ects of wall shear stress and liquid acceleration due to void changes that may in uence gas holdup in cocurrent bubble columns (Hills, 1976; Merchuk, 1981; Kumar et al., 1996); however, these e ects were estimated to be negligible for the conditions of this study. The overall gas holdup is de ned as  = (1 + 2 + 3 )=3, the average gas holdup in the three lower sections. When “gas holdup” is mentioned alone, it refers to the overall gas holdup. 3. Results 3.1. Observations of the time-dependent gas holdup variation Fig. 3 shows the time-dependent gas holdup variation in a cocurrent bubble column operated using Procedure I with Ul = 8 cm=s. Results are presented for both sparger orientations, and the date the data were acquired is provided in the legend. The experiments were started separately for each sparger orientation with fresh tap water provided from the same source. The dashed (solid) lines indicate the data obtained with the sparger holes facing upward (downward). The gas holdup obtained with the sparger facing upward is higher than that obtained with the sparger facing downward. However, in both cases, the gas holdup decreases from day to day; this variation is small when the sparger faces upward but rather signi cant when facing downward. The time-dependent gas holdup variation is more signi cant at higher super cial gas velocities. Steady state is obtained with Procedure I at both sparger orientations if enough experimental runs are carried out. Similar results are obtained when Ul is xed at 4 cm=s or slightly above 0 cm=s. Fig. 4 compares the gas holdup variation with time in cocurrent ows using Procedure II and both sparger orientations. In both cases, experiments are started with fresh tap water and Ug = 20 cm=s and Ul = 8 cm=s. The gas holdup is higher when the sparger faces upward. Variation in gas holdup is negligible when the sparger faces upward, but signi cant when the sparger faces downward. Assuming a rst-order time response, the time constant for the downward facing results is ∼155 min. When Ug and Ul are xed at other values in the range 0 –20 and 0 –10 cm=s, respectively, similar time-dependent behavior of gas holdup is observed. C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 627 Fig. 4. Time-dependent gas holdup variation in the cocurrent bubble column obtained via Procedure II. 3.2. E ects of water quality Fig. 5. Time-dependent gas holdup variation in the column operated in semi-batch mode obtained via Procedure II. The gas holdup variation at both sparger orientations in the column operated in semi-batch mode is presented in Fig. 5 with Ug = 20 cm=s. Gas holdup decreases signi cantly with time for both sparger orientations, and no distinct di erence is observed between the gas holdups for the two cases. Steady state for both sparger orientations is reached in ∼145 min. As shown in Table 1, the time constant is in uenced by experimental conditions, with Ug being the most signi cant. In general, the gas holdup variation time constant decreases with increasing Ug . To identify the mechanisms behind the complex time-dependent gas holdup behavior in the air–water bubble column, additional experiments were carried out. The additional tests focused on the water because no oil was observed or smelled at the exit of the air lter, as indicated by Ueyama et al. (1989). Water properties including temperature, surface tension, pH, electrical conductivity, total dissolved solids, and kinematic viscosity were measured in Procedures I and II. No signi cant variation of surface tension, electrical conductivity, total dissolved solids, or kinematic viscosity is identied. Temperature increases slightly with time in Procedure II, due to energy dissipation in the pump. Each time an experiment is initiated with fresh tap water, using Procedure I or II, pH decreases from about 9.3 to 8.4 during the rst 30 min and then is relatively constant with time, while the gas holdup still decreases with time. The decrease in pH is the result of carbon dioxide dissolution in water (Benjamin, 2002). Recording the above water properties every 30 min for several hours reveals that none of the above properties are the main cause of the time-dependent gas holdup variation. Fig. 6 compares gas holdup versus time using 2 di erent types of water with the sparger facing downward and Ug = 20 cm=s and Ul = 8 cm=s. The open symbols represent gas holdup taken using fresh tap water. The solid symbols indicate the gas holdup obtained using “used” or steady-state water; this water is conditioned fresh tap water using Procedure II with Ul = 8 cm=s and Ug = 9 cm=s until a steady-state gas holdup is obtained. There is no signi cant time-dependent gas holdup variation in the used water, while the fresh tap water has a distinct gas holdup 628 C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 Table 1 Selected experimental conditions and time constants Cases a b c d e Experimental conditions Column mode Sparger orientation Ul (cm=s) Ug (cm=s) Overall gas holdup range (Section 2 for cases d and e) Cocurrent Cocurrent Cocurrent Semi-batch Semi-batch Downward Downward Downward Downward Upward 8 8 4 NA NA 20 9 8 20 20 0.1962– 0.1767 0.1267– 0.1092 0.1176 – 0.1096 0.2216 – 0.1932 0.2246 – 0.2009 Fig. 6. Comparison of the time-dependent variations of gas holdup in fresh tap water and “steady-state” water. time-dependence with a time constant of ∼155 min. Thus, apparently something in the water causes the time-dependent gas holdup variation. Fig. 7 shows additional semi-batch mode data with the sparger facing up and Ug = 20 cm=s, and reveals the time-dependent gas holdup behavior in fresh tap water (unprocessed water) and water resulting from mixing the fresh tap water in the reservoir for 150 min (processed water). Note the mixing operation only involved operating the reservoir mixer (Fig. 1) and no air was bubbled through the water. The gas holdup variation with time is signi cant in the unprocessed water, which is also shown in Fig. 5. In the processed water, the gas holdup is approximately steady with time and equals the steady-state value obtained in the unprocessed water. This con rms the role of water quality inferred from Fig. 6. Furthermore, it demonstrates that system operation changes the water quality and the water quality change is unlikely due to accumulation of impurities from the air. In Fig. 8, the signi cance of the time-dependence of gas holdup at low and high Ug in the semi-batch mode bubble Time constant (min) 155 350 220 145 150 Fig. 7. Comparison of time-dependent gas holdup variations in processed water and unprocessed water in the bubble column operated in semi-batch mode. column is compared with the sparger facing upward. The solid triangles represent the gas holdup in fresh tap water at Ug = 4 cm=s; little variation with time is observed. The open triangles represent the gas holdup at Ug = 20 cm=s taken in the water just used at Ug = 4 cm=s. Two important points are revealed in this gure: (i) the time-dependent gas holdup variation is signi cant only at high Ug ; and (ii) the time-dependence of the gas holdup at higher Ug is signi cant even in water that was used to obtain a steady state at a low Ug . This shows that the degree of water quality change is a ected by the super cial gas velocity, and a larger change in water quality is observed at higher Ug . Fig. 9 shows the time-dependent gas holdup variation obtained during two consecutive periods separated by a 12-h idle period during which the system was stopped but the water was not changed. The experimental settings in the two periods are kept the same and the sparger faces downward. The solid symbols represent data obtained during the rst period and the open symbols represent the second. The 12-h idle period is omitted from the time axis. Fig. 9(a) C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 Fig. 8. Comparison of time-dependent gas holdup variations at low and high super cial gas velocities in the semi-batch mode bubble column. Fig. 9. Time-dependent variation of gas holdup during two consecutive time periods separated by an idle period; (a) semi-batch mode, (b) cocurrent mode. 629 shows the results obtained in the semi-batch mode with Ug = 19 cm=s. The gas holdup variation with time during the second period simply continues the trend from the rst period. The combined gas holdup versus time curve is very similar to those in Fig. 5, implying the water quality in the semi-batch mode column did not change signi cantly during the idle period. Fig. 9(b) presents the results in the cocurrent mode with Ug = 10 cm=s and Ul = 4 cm=s. The combined curve in Fig. 9(b) is also similar to that for a cocurrent ow with the sparger facing downward (e.g., Fig. 4), although there is a small step change between the curves during the two periods. Hence, a signi cant water quality change happens during the air–water contacting periods, as well as the idle period during cocurrent bubble column operation. It is possible that the gas holdup variation observed in the present study is due to microbubbles that exist in fresh tap water and dissipate over time. However, Gavrilov (1969) found that the gas holdup in tap water due to microbubbles is initially on the order of 10−9 and, after about 5 h, remains constant on the order of 10−10 . In the present study, the gas holdup variation is on the order of 0.01 (e.g., Fig. 4). Therefore, dissipation of entrained microbubbles in tap water are not the main cause of the time-dependent gas holdup variation. The change in gas holdup with respect to time is attributed to water quality variations. This agrees with the results of Anderson and Quinn (1970) and Ueyama et al. (1989). Although it is still not exactly clear how the tap water quality changes and how the change results in di erent gas holdups, it is however, plausible to attribute the changes to the presence of a VS, which may be present in the water in extremely small concentrations (Michalson, 1968; Gelover et al., 2000; Skjevrak et al., 2003). One important common characteristic of a VS is that it can signi cantly suppress bubble coalescence, even in extremely low concentrations, and the higher the concentrations, the stronger the inhibiting e ect on bubble coalescence (Jamialahmadi and Muller-Steinhagen, 1992; Krishna et al., 2000; Zahradnik et al., 1999; Camarasa et al., 1999; Ueyama et al., 1993). The VS evaporates from the water to the air during system operation. The higher the system agitation, the faster the evaporation rate and the lower the VS equilibrium concentration. During system operation, the VS gradually evaporates from the water, resulting in lower concentrations; this leads to the enhancement of the bubble coalescence as time progresses, which, in turn, causes the gas holdup to decrease with time (as shown in Figs. 3–5). When the VS concentration decreases to the equilibrium concentration, or becomes lower than the minimum required concentrations (Ueyama et al., 1993), a steady-state gas holdup is reached. The evaporation process is very slow, so a long time period is required to reach equilibrium. When “used” steady-state water is used to take gas holdup data (Fig. 6) at other operating conditions, the time-dependence of gas holdup is 630 C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 not signi cant because additional VS evaporation did not occur. When fresh tap water is agitated in the reservoir, the VS evaporates into the air. Over a long time period, the VS concentration in the water is reduced to be no higher than the equilibrium concentration corresponding to the agitation level in the bubble column under the given ow conditions. Consequently, when processed tap water is used, negligible VS evaporation exists and the time-dependence of gas holdup is insigni cant, as shown in Fig. 7. Once the VS evaporates from the water, it does not go back into solution, so the bubble coalescence inhibiting ability of the water does not increase during system idle periods. However, a small amount of the VS may still evaporate from the water surface during idle periods, albeit at a slower rate, and after a long time period, could possibly enhance bubble coalescence and result in a sensible decrease in the gas holdup, as shown in Fig. 9(b). This is not observed in semi-batch operation (Fig. 9(a)) because the steady-state has already been reached during the rst day and no additional VS evaporates during the idle period. In addition to system agitation (i.e., mixing), VS evaporation can be enhanced by heating; this can be used to explain the results of Anderson and Quinn (1970). The authors heated the tap water to 55◦ C and cooled it back to 15◦ C with a provision for resaturating the water with air. They found the gas holdup in this water was similar to distilled water. Other observations of Anderson and Quinn (1970) can also be explained by assuming VS evaporation in their system. 3.3. E ects of sparger orientation and super cial liquid velocity When the sparger faces downward, bubbles detach from the aeration holes in the opposite direction as that of the liquid ow. To detach from the aeration holes, the bubbles must overcome the resistance to the liquid ow. This in uences the bubble behavior in two ways. First, the bubbles su er a higher pressure from the aeration holes to overcome the liquid ow. This increases the bubble size. Furthermore, once the bubbles detach from the aeration holes, the pressure di erence between the inside and outside of the bubble reduces signi cantly because the liquid and bubbles now ow in the same direction, resulting in further bubble expansion. The increased bubble size also enhances bubble rise velocity. Second, when bubbles detach from the aeration holes, they must detour around the sparger arms and rise through the space between adjacent sparger arms. This space is narrow and must be shared by bubbles detaching from aeration holes on neighboring arms. Consequently, bubble coalescence in the aeration zone is greatly enhanced. The bubble coalescence enhancing e ect is stronger at higher Ul when the sparger faces downward. When the sparger faces upward, bubbles detach from the aeration holes in the same direction as that of the liquid ow. With the help of the owing liquid, newly generated bubbles are carried away from the aeration holes and rise directly upward without touching neighboring bubbles, and, through visual observations, these bubbles are smaller than those observed with the sparger facing downward. Both e ects reduce bubble coalescence in the aeration zone. The bubble coalescence inhibiting e ect is stronger at higher Ul . Hence, the e ects of liquid velocity on bubble coalescence with a downward facing sparger are completely opposite to that of an upward facing sparger; this results in a higher gas holdup when the sparger faces upward than when the sparger faces downward in cocurrent ows (Fig. 3). In the semi-batch operation mode, the average liquid velocity is zero in the aeration zone. On one hand, when the sparger faces upward, bubbles detach from the aeration holes in larger sizes and bubble coalescence is stronger in the aeration zone than in cocurrent ows at high Ug (according to visual observations in the semi-batch mode, at high Ug , many of the bubbles went down to the space between the sparger arms due to backmixing, which is less likely to happen in cocurrent ow when the sparger faces upward). On the other hand, when the sparger faces downward, the detached bubbles are smaller than in cocurrent operation because of negligible liquid ow. Although the bubbles must still rise through the space between sparger arms, coalescence is less severe than in cocurrent operation when the sparger faces downward because of the initial small bubble sizes. Hence, the di erence in bubble coalescence tendency for both sparger orientations is small in the semi-batch mode. As a result, the di erence between the time-dependence of gas holdup between the two orientations is also small (Fig. 5). Ueyama et al. (1993) observed that the coalescence time of two bubbles consist of two parts. One was the contacting time period and the other was the initial stage of lm rupture. The addition of n-alcohol or surfactant in a concentration higher than a minimum value considerably increased the rst part, but the second was constant at about 0:3 ms regardless of concentration. The e ect of surfactants on coalescence time agrees with the observations of Tse et al. (1998). Ueyama et al. (1993) also found that the age of bubbles signi cantly in uenced bubble coalescence, with the youngest bubbles coalescing much faster than older bubbles due to the accumulation of surfactants at the gas–liquid interface, which was a slower process compared to bubble coalescence. In our experimental facility, bubbles are young near the aeration zone, so the required coalescence time is short. When the VS concentration decreases, the required coalescence time is reduced. Because most bubble–bubble contacts are very short, the enhancement to bubble coalescence is signi cant, provided that the bubble contacting frequency is large. In the upper region of the bubble C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 column, bubbles are older, so the required coalescence time is longer. Although the total bubble contacting frequency increases due to enhanced bubble-induced turbulence, the portion of long-duration bubble contacts is small. When the VS concentration decreases, the required bubble coalescence time is still large, and the VS concentration change results in a negligibly enhancement of bubble coalescence. Thus, the time-dependent gas holdup variation due to local VS concentration change is signi cant only in the bottom of the bubble column. However, in cocurrent operation mode with the sparger facing upward, there are not many bubble–bubble contacts in the lower column region, so the time-dependent gas holdup variation is insigni cant over the entire bubble column. When the sparger faces downward, bubble contacting in the aeration zone is very frequent, and the time-dependence of gas holdup is signi cant (Fig. 4). 3.4. E ects of super cial gas velocity In the aeration zone, the size and frequency of newly generated bubbles increase with increasing super cial gas velocity (Park et al., 1977; Terasaka et al., 1999; Snabre and Magnifotcham, 1998). At high Ug , the opportunity of bubble coalescence is larger than at low Ug . Hence, VS concentration change has a more signi cant e ect on gas holdup at high Ug . However, in the cocurrent mode this e ect of Ug is weakened by Ul , which reduces the bubble size and backmixing in the aeration zone. Furthermore, in a gas–liquid bubble column, the turbulence intensity increases with increasing Ug and decreasing Ul (Michelsen and Ostergaard, 1970; Eissa et al., 1971). The evaporation rate also increases with increasing turbulence intensity (Dewulf et al., 1998; Seno et al., 1990). Thus at higher Ug , as shown in Table 1, the time constants are smaller with increasing Ug due to faster VS evaporation, with all other conditions xed. Additionally, the equilibrium VS concentration in water is higher at a lower turbulence level. Therefore, a steady-state can be reached at a low Ug (low turbulence) using fresh tap water and Procedure II. Using this water at a higher Ug (higher turbulence), time-dependent gas holdup variation may be significant (Fig. 8) or insigni cant (Fig. 6), depending on how large a di erence between the equilibrium VS concentrations corresponding to the respective super cial gas velocities. However, if fresh tap water is at rst used at a high Ug and a steady state is reached, then no signi cant time-dependent variation is observed using the same water at a low Ug . In a cocurrent bubble column, the liquid velocity reduces the relative velocity between the liquid and gas and hence, the bubble-induced turbulence intensity, resulting in lower VS evaporation rates. Furthermore, the cocurrent operation mode involves about 6 times as much water than the 631 semi-batch mode. Both factors make the VS concentration decrease at a much slower rate in a cocurrent system than in a semi-batch system. 4. Conclusions Time-dependent variation of gas holdup was observed in an air–water cocurrent bubble column, which was also operated in semi-batch mode. Experimental results showed that the time-dependent gas holdup variation was related to water quality, column operation mode, sparger orientation, and super cial gas and liquid velocity. The gas holdup variation with time was attributed to a coalescence-inhibiting VS, which existed in the tap water at very low concentrations. As the VS evaporated from the water, the coalescence-inhibiting ability of the water was continuously reduced; consequently, the gas holdup in the bubble column changed continuously with time. The sparger orientation strongly in uenced bubble size and coalescence in the aeration zone, especially when the super cial liquid velocity was above zero. Coalescence in the aeration zone was much more signi cant when the sparger faced downward when compared to the upward orientation; this resulted in the gas holdup variation that was more signi cant when the sparger faced downward in cocurrent ows. In semi-batch operation, the e ect of sparger orientation was insigni cant because of modi ed hydrodynamics and competing bubble–bubble interaction e ects. The super cial gas velocity in uenced bubble size and frequency in the aeration zone, as well as the bubble-induced turbulence intensity, which was also a function of super cial liquid velocity and the attendant experimental conditions. The bubble-induced turbulence intensity a ected the VS evaporation rate. Two important di erences between cocurrent and semi-batch operation were (i) the cocurrent mode involved a much larger volume of water, making the rate of concentration change of VS much slower than in the semi-batch mode; and (ii) in the cocurrent mode, Ul reduced bubble-induced turbulence intensity and bubble size in the aeration zone. Thus the signi cance of the time-dependence of gas holdup was a strong function of column operation mode. This work is useful to understand the importance of experimental reproducibility of gas–liquid bubble column data. It also furthers our understanding to how a small change in water quality results in a signi cant gas holdup variation. Although preliminary water quality tests for volatile organic compounds (VOCs) revealed some changes with time, de nitive conclusions could not be drawn from the results and further work is necessary to identify the species and concentration of the volatile substance in the tap water that played a critical role in the gas holdup variation. 632 C. Tang, T.J. Heindel / Chemical Engineering Science 59 (2004) 623 – 632 Notation Ug Ul p super cial gas velocity, cm=s super cial liquid velocity, cm=s pressure, Pa Greek letters
 di erence gas holdup Subscripts l h i 0 at the lower end at the higher end column section number measured in static water Acknowledgements Portions of this work were supported by the USDA under agreement number 2110-35103-11259. References Anderson, J.L., Quinn, J.A., 1970. Bubble columns: ow transitions in the presence of trace contaminants. Chemical Engineering Science 25, 373–380. Benjamin, M.M., 2002. Water Chemistry. McGraw-Hill, New York. Camarasa, E., Vial, C., Poncin, S., Wild, G., Midoux, N., Bouillard, J., 1999. In uence of coalescence behaviour of the liquid and of gas sparging on hydrodynamics and bubble characteristics in a bubble column. Chemical Engineering and Processing 38, 329–344. Dewulf, J., Van Langenhove, H., Heireman, B., 1998. Air/water exchange of volatile organic compounds from waters in the transient and turbulent regime. Water Research 32, 2106–2112. Eissa, S.H., Ei-Halwagi, M.M., Saleh, M.A., 1971. Axial and radial mixing in a cocurrent bubble column. Industrial Engineering Chemistry Process Design and Development 10, 31–36. Gavrilov, L.R., 1969. On the size distribution of gas bubbles in water. Soviet Physics—Acoustics 15, 22–24. Gelover, S., Bandala, E.R., Leal-Ascencio, T., Perez, S., Martinez, E., 2000. GC-MS determination of volatile organic compounds in drinking water supplies in Mexico. Environmental Toxicology 15 (2), 131–139. Hills, J.H., 1976. The operation of a bubble column at high throughputs. Chemical Engineering Journal 12, 89–99. Jamialahmadi, M.C., Muller-Steinhagen, H., 1992. E ect of alcohol, organic acid and potassium chloride concentration on bubble size, bubble rise velocity and gas hold-up in bubble columns. Chemical Engineering Journal and the Biochemical Engineering Journal 50, 47–56. Kelkar, B.G., Phulgaonkar, S.R., Shah, Y., 1983. E ect of electrolyte solutions on hydrodynamic and backmixing characteristics in bubble columns. Chemical Engineering Journal and the Biochemical Engineering Journal 27, 125–133. Krishna, R., Urseanu, M.I., Dreher, A.J., 2000. Gas hold-up in bubble columns: in uence of alcohol addition versus operation at elevated pressures. Chemical Engineering and Processing 39, 371–378. Kumar, S.B., Dudukovic, M.P., Toseland, B.A., 1996. Measurement techniques for local and global uid dynamic quantities in two and three phase systems. In: Chaouki, J., Larachi, F., Dudukovic, M.P. (Eds.), Non-Invasive Monitoring of Multiphase Flows. Elsevier, New York, pp. 1–45. Maruyama, T., Yoshida, S., Mizushina, T., 1981. The ow transition in a bubble column. Journal of Chemical Engineering of Japan 14, 352–357. Merchuk, J.C., 1981. Local hold-up and liquid velocity in air-lift reactors. A.I.Ch.E. Journal 27, 377–388. Michalson, A.W., 1968. High quality water via ion exchange. Chemical Engineering Progress 64 (10), 67–73. Michelsen, M.L., Ostergaard, K., 1970. Hold-up and uid mixing in gas-liquid uidized beds. The Chemical Engineering Journal 1, 37–45. Miyahara, T., Matsuba, Y., Takahashi, T., 1983. Size of bubbles generated from perforated plates. International Chemical Engineering 23, 517–523. Park, Y., Tyler, A.L., de Nevers, N., 1977. Chamber ori ce interaction in the formation of bubbles. Chemical Engineering Science 32, 907–916. Seno, T., Uchida, S., Tsuyutani, S., 1990. Mass transfer in countercurrent and cocurrent bubble columns. Chemical Engineering & Technology 13, 113–118. Skjevrak, I., Due, A., Gjerstad, K.O., Herikstad, H., 2003. Volatile organic components migrating from plastic pipes (HDPE, PEX and PVC) into drinking water. Water Research 37, 1912–1920. Snabre, P., Magnifotcham, F., 1998. Formation and rise of a bubble stream in a viscous liquid. The European Physical Journal B 4, 369–377. Terasaka, K., Tsuge, H., Matsue, H., 1999. Bubble formation in cocurrently upward owing liquid. Canadian Journal of Chemical Engineering 77, 458–464. Tse, K., Martin, T., McFarlane, C.M., Nienow, A.W., 1998. Visualisation of bubble coalescence in a coalescence cell, a stirred tank and a bubble column. Chemical Engineering Science 53, 4031–4036. Ueyama, K., Tsuru, T., Furusaki, S., 1989. Flow transition in a bubble column. International Chemical Engineering 29, 523–529. Ueyama, K., Saeki, M., Matsukata, M., 1993. Development of system for measuring bubble coalescence time by a laser. Journal of Chemical Engineering of Japan 26, 308–314. Zahradnik, J., Kuncova, G., Fialova, M., 1999. The e ect of surface active additives on bubble coalescence and gas holdup in viscous aerated batches. Chemical Engineering Science 54, 2401–2408.