- Email: [email protected]
Modeling and scale-up simulation of U-tube ozone oxidation reactor for treating drinking water
Chemical Engineering Science 60 (2005) 6360 – 6370 www.elsevier.com/locate/ces
Modeling and scale-up simulation of U-tube ozone oxidation reactor for treating drinking water Katsuhiko Muroyamaa,∗ , Masahiro Yamasakia , Mamoru Shimizua , Eiji Shibutania , Takeshi Tsujib a Department of Chemical Engineering, Kansai University, Osaka 564-8680, Japan b Aqua Technology Research Dept., JFE Engineering Corporation, Kawasaki City, Kanagawa 210-0855, Japan
Received 1 November 2004; received in revised form 20 April 2005; accepted 20 April 2005 Available online 20 June 2005
Abstract In the present study, we developed a novel simulation model of the U-tube reactor for treating drinking water, which is composed of a coaxial inner tube serving as an efficient concurrent down-flow ozone dissolver and an outer column carrying out reactions between ozone and organic substances including odorous materials (2-methylisoborneol: 2-MIB) dissolved in the raw water. We assume that the U-tube is composed of a plug flow section (inner tube) followed by a tanks-in-series section (outer bubble column) and take into account the effect of the hydrostatic pressurization on the flow and absorption equilibrium for the gaseous components including ozone and other inactive species in developing the mass balance models. An algorithm is constructed of the differential multiple mass balance equations for the inner tube sections and multiple difference mass balance equations in the series tanks in the outer column section to enable the scale-up from a pilot plant to a full-scale plant. The gas holdup and gas–liquid mass transfer coefficient were measured in a model reactor and correlated for the use of the simulation calculation. Available literature data and correlations on the rates of reactions between ozone and organic substances including odorous material 2-MIB, gas–liquid equilibrium for active and inactive gases and axial fluid mixing properties are also incorporated in the simulation calculation. The simulation results well explained the available data of the ozone absorption efficiency and the removal efficiency of the odorous material in a pilot U-tube reactor. The simulation procedure was also successfully extended to verify the performance of a full-scale U-tube reactor. It is shown that the ozone absorption is practically a single function of the gas/liquid ratio while the removal efficiency of the odorous material is a single function of the ozone dose for a specified U-tube configuration. 䉷 2005 Elsevier Ltd. All rights reserved. Keywords: Drinking water; Ozone treatment; U-tube reactor; Simulation; Scale-up; Volumetric gas–liquid mass transfer coefficient; Gas holdup
1. Introduction Due to utrophication the increased inclusion of nitric compounds and humic substances occurs in the river water, inevitably resulting in an excess chlorination which in turn generates unwanted chlorine compounds such as trihalomethanes in the treated water. In an advanced drinking water treatment, the employment of an ozone treatment followed by an activated carbon adsorption is becoming popular in Japan (Somiya, 1989). The ozone treatment can reduce the odorous materials such as geosmin and 2-methylisoborneol (2-MIB) and the precursors to the ∗ Corresponding author. Tel.: +81 6 6368 0945; fax: +81 6 6388 8869.
E-mail address: [email protected] (K. Muroyama). 0009-2509/$ - see front matter 䉷 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2005.04.064
trihalomethanes, and compensate for the chlorination treatment, thus reducing the occurrence of the total chlorinated organic compounds. Note that ozone is an environmental risk free substance because the dissolved ozone in water readily self-decomposes and reduces to oxygen (Sato, 1992). U-tube ozonation reactor is a novel bubble column constructed of a concentric downflow tube and a outer column; in the downflow tube ozonated air or ozonated oxygen is mixed with fresh river water and an efficient dissolution of ozone is achieved under hydrostatical pressurization, while in the outer column the dissolved ozone reacts with the precursors to trihalomethanes and the odorous materials such as geosmin and 2-MIB for a sufficient residence time to reduce the concentrations of these target materials within
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
an allowable level. Hydrodynamics of the U-tube reactor was studied on the flow pattern and the gas holdup in the concurrent downflow tube was measured and correlated by Roustan et al. (1990, 1992a,b). The liquid mixing properties in a full scale U-tube reactor were also studied by Roustan et al. (1993). In our previous study (Muroyama et al., 1999), we proposed the design model of a U-tube ozone oxidation reactor for treating drinking water, assuming that in the inner tube, the flows of gas and liquid are both in plug flow mode and in the outer tube, the flow of the gas is still in plug flow mode while the flow of the liquid can be modeled by tanks in series. The proposed model well simulated a pilot U-tube reactor, verifying that the calculated results well predicted the data on the ozone absorption efficiency and the removal efficiency of odorous material 2-MIB. In the present study, we developed a novel design model for the U-tube reactor considering the effect of hydrostatical pressurization on the volumetric flow and the gas–liquid equilibriums for gaseous components including ozone and other inactive species. The ozone absorption, the reaction between dissolved ozone and dissolved odorous material 2MIB in the liquid phase, the flow and mixing of both gas and liquid phases are combined to set up the multiple differential mass balance equations in the inner tube and the multiple mass balance difference equations based on the tanks in series model in the outer column. The physical absorption of inactive gases is also considered in the mass balance equations. An algorithm is constructed of the multiple mass balance equations with appropriate boundary conditions in the two reactor sections and solved to simulate the reactor performance characteristics including the ozone absorption efficiency and the decomposition efficiency of the odorous material for the U-tube ozonation reactor treating drinking water. Available data on the reaction kinetics, gas–liquid equilibrium, absorption rates of gases and fluid mixing reported in the literature or experimentally obtained by ourselves are also incorporated in the simulation calculations. The reactor performance characteristics evaluated for the pilot plant are well predicted by the simulation calculations. The simulation model is also successfully extended to verify the reactor performance for a full-scale U-tube reactor treating drinking water.
2. Experimental 2.1. Experimental apparatus Fig. 1 shows the experimental U-tube apparatus for measuring the hydrodynamic properties and the gas–liquid mass transfer characteristics. The dimensions of the experimental column are as follows; the outer column is 454 mm in I.D. and 3550 mm in height and the inner tube is 75 mm in I.D. and its end is opened at 100 mm above the outer column bottom. The inner tube is fitted coaxially in the outer col-
6361
1 4
6
5
2 3
1 Main column 2 liquid reservoir
4 Liquid flow meter 5 Air supply
3 Liquid pump
6 Gas flow meter
Fig. 1. Experimental apparatus for measuring hydrodynamic and mass transfer characteristics.
Table 1 Operating conditions for experimental system for measuring hydrodynamics and gas–liquid mass transfer characteristics Inner tube Outer column
Superficial Superficial Superficial Superficial
liquid velocity (m/s) gas velocity (m/s) liquid velocity (m/s) gas velocity (m/s)
0.565–1.885 0.0113–0.339 0.0212–0.0739 0.00443–0.0133
umn and the length of its straight portion well exceeds the length of the outer column. The experimental conditions for both phase superficial velocities are listed in Table 1. Two types of gas distributor were used; the first one is a single pipe of 6 mm I.D. stainless steel pipe and the second one is a four-way porous-tips nozzle, welded with sintered porous tips with average pore diameter of 40 m. 2.2. Gas holdup measurement The axial distribution of gas holdup in the inner tube was measured using pairs of electro conductivity cells of thin 25 mm × 60 mm square stainless steel plates which were mounted facing each other on the opposite sides of the inner wall surface at seven axial positions. After a steady state condition for the gas–liquid downflow was established in the inner tube, the inter-cell impedance was measured with 2 kHz AC by using a LCR meter (KC-547, KOKUYO Electric). In advance a correlation between the liquid holdup and the ratio of the inter-cell impedance for the single liquid flow to that for the gas–liquid flow or for the liquid–solid fluidization was obtained. The gas holdup was measured also in the outer column using a static pressure gradient method.
6362
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
2.3. Volumetric mass transfer coefficient, kL a, in the inner tube
G [-]
10-1
nozzle1 nozzle2 UL [m/s]
10-2
0.75 1.13 nozzle1:Single pipe
1.51
nozzle2:Porous -tipsnozzle
1.89
10-3 10-2
10-1 UG [m/s]
10-1
Normal Bubble column Akita-Yoshida(1973)
10-2
10-3 nozzle1:Single pipe nozzle2:Porous -tips nozzle
3. Experimental results for gas holdup and mass transfer
10-4 10-4
10-3
nozzle1 nozzle2 UL [m/s] 0.0296 0.0433 0.0591 0.0793 10-2
10-1
100
UG [m/s]
3.1. Gas holdup The gas holdup, g in the inner tube is plotted against the superficial gas velocity in Fig. 2. The value of g in the concurrent downflow tube increases with superficial gas velocity while it significantly decreases with increasing downward liquid velocity. Note that it is significantly lower than that in the normal bubble column as typically represented by a solid line for Akita–Yoshida’s correlation. This may be explained as follows; in the high shear condition at a high downward liquid velocity the bubbles are finely divided and homogeneously dispersed in the liquid flow. The increase of liquid flow rate steadily increases the downward velocity of gas bubbles and then the gas holdup decreases with increasing liquid velocity. Note that the values of gas holdup for the single pipe nozzle are significantly greater than those for the four-way porous-tips nozzle since the former nozzle creates course bubbles with a wide size distribution while the latter nozzle causes fine bubbles with a narrow distribution. A regression analysis leads to the following empirical gas holdup correlations for the two gas distributors: 0.998 −1.115 gi = 0.821Ugi ULi
100
Fig. 2. Variation of G versus UG in the inner tube.
G [-]
Volumetric mass transfer coefficient in the inner tube was measured by physical absorption of pure oxygen; pure oxygen gas was introduced at a controlled rate through the gas nozzle at the top of the inner tube under a controlled liquid flow, while nitrogen gas was bubbled through a ring-shaped sparger fitted at the bottom of the outer column at a sufficiently high rate to desorb oxygen from the water in the outer column. After a steady state was established, the dissolved oxygen concentrations at the inlet and outlet of the inner tube were measured simultaneously by means of DO meters (YAI Model 57, YSI Japan). The axial differential mass balance equations are formulated to solve the oxygen concentrations in the liquid and gas phases based on the assumption that the gas and liquid flow downward in plug flow modes and the value of the mass transfer coefficient kL a is a constant along the entire inner tube length, while the effect of hydrostatic pressure at any axial position on the volumetric gas flow rate and the gas–liquid equilibrium is considered. The multiple first order differential equations were numerically solved to obtain an optimal value of kL a by an iterative procedure by fitting the calculated outlet dissolved oxygen concentration with the observed one. The mass balance equations and the iterative procedures to solve their equations are the same as those described for the absorption of inactive gases in the later section and omitted here.
Normal Bubble column Akita-Yoshida(1973)
(single-pipe nozzle),
(1)
Fig. 3. Variation of G versus UG in the outer column. 1.066 −1.153 gi = 0.731Ugi ULi
(porous-tips nozzle).
(2)
Fig. 3 shows the plot of the gas holdup values in the outer column against the superficial gas velocity and compares the calculated value from the correlation by Akita and Yoshida (1973) for the normal bubble column. It is obvious that the experimental values are fairly well explained by the Akita–Yoshida’s correlation; however, the experimental gas holdup values are laying slightly below except in a range of lower gas velocities less than 3 × 10−3 m/s, where the experimental gas holdup values slightly exceed the calculated those from Akita–Yoshida’s equation. 3.2. Mass transfer characteristics in the inner tube Fig. 4 shows the plots of kL a value versus UG in the inner tube for two types of gas distributors indicating that the values of kL a increases with UG . The values of kL a in the inner tube are significantly greater than those in the nor-
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
Differential mass balance for the gas phase ozone is represented as dGO3 ∗ − CLO3 ). (6) = −(kL a)O3 (CLO MO3 3 dz
kLa [1/s]
10-1 Normal Bubble column Akita-Yoshida(1973)
10-2
nozzle1: Single pipe nozzle2: Porous-tips nozzle
1.13 1.51
UL
1.89 10-2
10-1
CGO3 = GO3 MO3 /UG .
(7)
Mass balance equation for the liquid-phase inactive gas component (j; oxygen) is given by
0.75
10-3
Here, the liquid phase and gas phase ozone concentrations, ∗ CLO and CGO3 are defined as 3 ∗ = mO3 CGO3 , CLO 3
nozzle1 nozzle2 UL [m/s]
10-4
6363
100
UG [m/s]
dCLj ∗ = (kL a)j (CLj − CLj ). dz
Overall mass balance for the specified gaseous inactive component (j; oxygen) is given by Gj = Gj 0 − UL (CLj − CLj 0 )/Mj .
Fig. 4. Variation of kL a versus UG .
(8)
(9)
The saturated liquid phase concentration of the inactive gas, ∗ , is defined as CLj mal bubble column as typically represented by a curve of Akita–Yoshida’s correlation. The values of kL a are slightly greater for the four-way porous-tips nozzle than for the single pipe nozzle; this is because the bubbles evolved from the four-way porous-tips nozzle are smaller and uniform in size distribution resulting in the higher specific interfacial area. The empirical correlations were formulated for the two types of nozzles as follows: 0.808 Single nozzle: kL a = 0.563UL0.121 UG
Porous-tips nozzle: kL a
0.788 = 0.600UL−0.242 UG
(3) (4)
∗ = mj CGj , CLj
CGj = Gj Mj /UG .
(10)
The total molar flow rate is given by Gt = GO3 + Gj .
(11)
The mol fraction and partial pressure of gaseous species are defined as Gj yj = , pj = Pt yj . (12) Gt Static pressure, Pt (Pa) at any depth is represented as Pt = P0 + L (1 − G )gz.
(13)
Superficial gas velocity is defined as 4. Simulation and scale-up modeling of U-tube reactor Considering the differences in the liquid mixing behaviors between the inner tube section and the outer column section, we employed the plug flow model for the liquid flow in the inner tube section and the series in tanks model in the outer column section, while the gas phase remains to be in a plug flow state in both sections. Mass balance modeling and necessary parameter setting in the two reactor sections will be described for the particular U-tube operation with ozonated oxygen in the following.
UG = Gt RT /Pt
(14)
Boundary conditions at the inner tube inlet are given for the gaseous species as follows: z = 0,
GO3 = GO3 0 ,
Gj = Gj 0 ,
CGO3 = CGO3 0 ,
CLj = CLj 0 .
(15)
Differential mass balance for the ozone consuming substance is given by UL
dXs = −(1 − G )kd CLO3 Xs . dz
(16)
4.1. Mass balances in the inner tube section
Differential mass balance for the odorous material 2-MIB is given by
Differential mass balance for the liquid phase ozone is formulated as
UL
UL
dCLO3 ∗ − CLO3 ) − (1 − G ) = (kL a)O3 (CLO 3 dz 1.5 + kr CLO3 Xs ). × (ks CLO 3
dXO = −(1 − G )kO CLO3 XO . dz
(17)
Boundary conditions for the ozone consuming substance and odorous material are expressed as (5)
z = 0,
Xs = Xs0 ,
XO = XO0 .
(18)
6364
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
4.2. Mass balances in the outer column
Superficial gas velocity is defined by
Mass balances for gas phase and liquid phase in the ith continuous tank are described below. Mass balance for the liquid phase ozone is expressed as UL (CLO3 ,i − CLO3 ,i−1 ) = MO3 (GO3 ,i−1 − GO3 ,i ) 1.5 + kr CLO3 ,i XS,i ). − (1 − G )Li (ks CLO 3 ,i
dGO3 ∗ = −(kL a)O3 (CLO − CLO3 ,i ). 3 dz
(19)
(20)
∗ , Here, the liquid phase saturated ozone concentration, CLO 3 is given by the following relations: ∗ CLO 3
= mO3 CGO3 ,
CGO3 = GO3 MO3 /UG .
(30)
Material balance for ozone consuming substance is given by UL (XL,i−1 − XL,i ) = (1 − g )Li kd CLO XS,i .
(31)
Mass balance for odorous material 2-MIB is given by
Differential mass balance for the gas phase ozone is presented as MO3
UG = Gt RT /Pt .
(21)
UL (XO,i−1 − XO,i ) = (1 − g )Li ko CLO XO,i
(32)
from which, the concentration of odorous materials can be readily determined. 4.3. Gas–liquid equilibrium and mass transfer characteristics The dimensionless Henry’s law constant, mj [-] is defined as follows: RT L [Pa m3 /(mol K)][K][kg/m3 ] = Hj MH2 O [Pa/mol frac.][kg/mol] = [mol frac.],
Eq. (20) can be integrated with the boundary condition at the inlet of the tank as given by
mj =
z = 0, GO3 = GO3 ,i−1 .
where Henry’s law constants for the inactive gases Hj (Pa) was estimated from the correlations given in a handbook (Society of Chemical Engineering Japan, 1999). That for ozone-water system associated with self-decomposition of ozone was estimated from the correlation given by Miyahara et al. (1994). The volumetric mass transfer coefficients for the gaseous species except oxygen in the inner tube were estimated from the correlation obtained from the physical oxygen absorption, (kL a)O2 , as follows: (kL a)j DLj 0.5 = , (34) (kL a)O2 DLO2
(22)
Differential mass balances for the inactive gas component (j; oxygen) are expressed as Mj
dGj ∗ − CLj ,i ), = −(kL a)j (CLj dz
(23)
∗ is given by the following relations: where CLj ∗ CLj = mj CGj ,
CGj = Gj Mj /UG ,
(24)
where mj is dimensionless Henry’s law constant for the inactive gas component. Eq. (23) can be integrated with a boundary condition at the bottom of the ith tank as given by z = 0, Gi = Gj,i−1 .
(25)
From the net mass balance, the gas phase molar flow rate at the exit of ith tank is given by Gj,i = Gj,i−1 − UL (CLj ,i − CLj ,i−1 )/Mj .
(26)
Note that we can determine the outlet liquid phase concentrations of ith tank, CLj ,i which is a constant throughout the tank compare the value of Gj,i obtained by integrating Eq. (23) with that from Eq. (26) to determine. The total molar flow rate is given by Gj . (27) Gt = GO3 + The mol fraction and partial pressure of gaseous species are defined as yj =
Gj , Gt
pj = Pt yj .
(28)
Static pressure at any depth is expressed as Pt = P0 + L (1 − G )gz.
(29)
(33)
where the diffusivities of the gaseous species were estimated from the Wilke–Chang correlation (Wilke and Chang, 1955). The volumetric mass transfer coefficient in the outer column was estimated from the correlation by Akita and Yoshida (1973) because the bubble flow mode and the gas holdup behavior in the outer column were the same to those in the normal bubble column because of low liquid velocity. The gas holdup in the inner tube for the single pipe nozzle was calculated from the correlation presented in Section 2.1. The gas holdup in the outer column was estimated by the correlation presented by Akita and Yoshida (1973). 4.4. Mixing parameter in the outer tube The number of series tanks indicating extent of liquid mixing in the outer tube section was evaluated as follows. The axial dispersion coefficient, Dz , in the normal bubble column can be estimated by the correlation of Deckwer (1992) which covers a range of lower superficial gas velocities as exemplified in the outer column of the U-tube. The dimensionless standard deviation for the normalized residence time
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
distribution for the axial dispersion model can be related to the inverse of the number of mixing tanks, J for the series in tanks model as follows: 2P 2 1 2 − (1 − exp(−P e) = , (35) = 2 Pe Pe J tP where Pe is the Peclet number for the axial dispersion model, P e = UL L/Dz . From Eq. (35), a real number is obtained for a specified gas–liquid operating conditions. In the simulation calculation of the tanks in series model, an additional reactor corresponding to the decimals part of the real number, j, was connected to the last one of the equivalent volume reactors corresponding to the integer part of j. 4.5. Kinetic parameters for the ozone oxidation reactions The self decomposition rate can be approximated by the first to second order kinetics with respect to the dissolved ozone concentration in the neutral water as encountered for the drinking water treatment (Morooka et al., 1978; Morioka et al., 1991; Miyahara et al., 1994). It was found that the correlation equation for the kinetic constant proposed by Morioka et al. deviates from other two correlations in the range of higher ozone concentration. Thus we employed the correlation of the kinetic constant for the ozone selfdecomposition most recently presented by Miyahara et al. (1994) whose correlation is shown below: ks = (4.0 × 1011 + 9.6 × 1014 [OH− ]0.8 ) × exp(−9.6 × 103 /T ).
−1 −1
kd = 0.8333 (m kg 3
s
).
Simulation parameter
Pilot plant
Full-scale plant
Column height (m) Inner tube diameter (m) Outer column diameter (m) Gas flow rate (m3 /s) Liquid flow rate (m3 /s) Liquid temperature (◦ C) Ozone concentration (kg/m3 ) Concentration of ozone consuming substance (kg/m3 ) Residence time in inner tube (s) Residence time in outer column (s) Number of mixing tanks in outer column (number)
17.0 0.047 0.312 1.04×10−4 5.36×10−3 20 0.148 1.60×10−3
27.2 0.77 3.95 1.89×10−2 0.944 20 0.148 1.60×10−3
5.5 237 18.6
13.4 340 4.38
which the axial distributions of species involved in the ozone oxidation process, such as, liquid phase ozone concentration, gas phase, and liquid phase concentration of odorous materials are numerically calculated. Note that the operating conditions listed in Table 2 correspond to the case where the ozonated oxygen is used to treat drinking water. The water temperature in the calculation was assumed to be a constant value of 20 ◦ C in the following calculations. 4.7. Algorithm and calculation procedure
(37a) (37b)
First order rate constant for the reaction of ozone with dissolved odorous material 2-MIB was correlated by Morioka et al. (1991) as follows: ko = exp(45.0 + 0.9[pH]) exp(−11100/T ).
Table 2 Reactor dimensions and operating conditions for simulation calculation
(36)
Note that Miyahara et al. determined that the ozone decomposition rate is proportional to 1.5th order with respect to the dissolved ozone concentration. Moniwa et al. (1991) investigated the first order reaction rate constant for the ozone consumption rate constant and the decomposition rate constant for the reaction of ozone with dissolved organic materials as follows: kr = 0.6667 (m3 kg−1 s−1 ),
6365
(38)
Note that Morioka et al. (1993) investigated the kinetics of ozone decomposition for geosmin and 2-MIB under the presence of carbonate, free chlorine, alcohols and volatile fatty acids. 4.6. Operating conditions The major dimensions, and operating conditions for the pilot and full-scale U-tube reactors are listed in Table 2, for
In the inner tube section, the axially differential equations for liquid-phase and gas-phase components including ozone, inactive gaseous components, and dissolved ozone consuming substances were solved by the Runge–Kutta–Gill methods based on the boundary conditions at the inlet of the inner tube for both phases. After determining the axial concentration distribution of these components, the liquidphase concentration of odorous material was solved. In the outer section, the difference mass balance equations for each mixed tank were solved by a trial-and-error procedure to fit the outlet liquid phase concentration with that from the overall mass balance between the phases. After determining the liquid phase concentrations, the concentration of odorous material was determined. The set of calculations were successively carried out to the final tank. The algorithm was constructed based on a C++ -language source program and all the calculations were carried out on a PC.
5. Simulation results In the following, the simulation calculations are carried out for the U-tube reactor operated with ozonated oxygen since the pilot and full-scale plants has been successfully demonstrated with ozonated oxygen.
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
outer column
equilibrium ozone conc. CLO3*=m×CGO3
0.03
Liq.-phase ozone conc. CLO3
kLa-25%
0.02 kLa0%
Pilot plant
0.01 kLa+25%
inner tube 0.003
1
kLa+25%
0.002
kLa0% kLa-25%
10
0
17
Pilot plant
0
0
inner tube
outer column
equilibrium ozone conc. CLO3*=m×CGO3
0.03
Liq.-phase ozone conc. CLO3
kLa-25%
0.02
kLa0%
Commercial plant
0.01 (b)
kLa+25%
0
0
10
20
27.2
20
10
0
10
17
10
[×10-7]
0
Axial depth [m]
(a) Liq.-phase ozone conc. [kg/m3]
0.04 Liq.-phase ozone conc. [kg/m3]
0
Axial depth [m]
(a)
(b)
10
0.5
0.001
0
0
1.5
outer column
Concentration of 2-MIB [kg/m3]
inner tube
Liq-phase ozone conc. [kg/m3]
Liq.-phase ozone conc. [kg/m3]
0.04
inner tube
-7 1.5 [×10 ]
outer column
kLa+25%
0.003
1 kLa0%
0.002
kLa-25%
0.5
0.001 Commercial plant
0
0
10
20 27.2 20 Axial depth [m]
10
0
0
concentration of 2-MIB [kg/m3]
6366
Fig. 6. Variation of dissolved ozone and odorous material 2-MIB for pilot and full-scale U-tube reactors.
Axial depth [m]
Fig. 5. Variation of dissolved ozone concentration with axial height for pilot and full-scale U-tube reactors.
5.1. Axial distribution of gaseous and liquid-phase components Fig. 5 shows the variations of the liquid phase and gas phase ozone concentrations versus axial depth in the inner tube and outer column sections. Note that the equilibrium liquid phase ozone concentration which is the value of product of dimensionless equilibrium constant, m, and the gas phase ozone concentration, m × CGO3 , is shown in place of the real gas phase ozone concentration to indicate how the liquid phase ozone concentration approaches their equilibrium value along the axial depth. In the figure kL a-0% means the value of the kL a estimated by the correlation for the single nozzle obtained in this work. Note that the value of kL a + 25% is increased by 25%, and that of kL a − 25% is decreased by 25% from that of kL a − 0%. It is obvious that in both reactors the dissolved ozone concentration increases with increasing axial depth in the inner tube while in the outer column it only slightly decreases with decreasing axial depth. In the outer column of the pilot plant the dissolved ozone concentration approaches their
equilibrium value and eventually exceeds the latter in the top section of the column where the ozone tends to be removed from the liquid into the gas. In the full-scale commercial plant, the dissolved ozone concentration approaches their equilibrium value at an axial depth of about 20 m and afterwards it becomes insensitive to the axial depth in the inner tube, while in the outer column it slightly decreases with decreasing axial depth and eventually exceeds the equilibrium concentration in the upper half region of the column where the desorption of ozone from liquid to gas surely occurs. The increase in kL a value should enhance the dissolution of ozone but the increase in the liquid phase ozone concentration significantly occurs only within an initial 10 m of the axial depth. Note that in the full-scale deep reactor, the influence of ±25% change in kL a on the dissolved ozone concentration is small because of the increased hydrostatic pressurization. Fig. 6 shows the variation of liquid-phase odorous material concentration versus axial depth together with that of liquid-phase ozone concentration. Because of its small outer column diameter, the number of mixing tank in series for the pilot plant is as high as 18.9 and the axial distribution of liquid-phase odorous material concentration is almost constant in the inner tube but it rather continuously decreases with increasing rise of axial depth. On the other hand in the full-scale reactor with a larger diameter of 3.95 m the num-
6
Liquid-phase equilibrium concentration
Commercial plant
5 4
0.1 kLa-25%
3
[ kk a0% a+25% L L
0.05
kLa+25%
2
kLa0%
1 kLa-25%
0
0.02 Volumetric gas velocity [m3/s]
outer column
inner tube
0.15
Gas-phase oxygen conc. [kg/m3]
Liquid-phase oxygen conc. [kg/m3]
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
inner tube
10
20 27.2 20 Axial depth [m]
10
0
Fig. 7. Axial oxygen concentration distributions in the liquid and gas phases for a full-scale U-tube reactor.
ber of the mixing tanks is only 4.4 and the concentration of the liquid-phase odorous material decreases in stepwise while the concentration drop in the first tank being largest. The liquid-phase odorous material concentration in the outer column only slightly varies with the ±25% changes from a kL a value estimated for the specified operating condition. Note that to improve the accuracy in the calculation model for the outer column the series tanks are considered to be composed of not only the number of tanks with an equal volume, corresponding to the integer part of the real number, J, derived from Eq. (35), but also an additional one of a smaller volume equivalent to the decimal fraction of J. 5.2. Axial distribution of inactive gas In Fig. 7, axial distributions of gas-phase and liquidphase concentrations of the inactive gas, oxygen, are shown for the specified conditions in the full-scale U-tube reactor. The gas-phase oxygen concentrations linearly increase with axial depth in the inner tube, and inversely linearly decrease with axial rise in the outer column due to the hydrostatic effect. The axial variations of the gas-phase oxygen concentration are almost neglected even for the ±25%-perturbed kL a values. The values of the liquid phase oxygen concentration also linearly increase with increasing axial depth in the inner tube, while in the outer tube, they remain almost constant values corresponding to the different kL a values. Fig. 8 shows the axial variations of the volumetric flow rate of gas, mainly consisted of oxygen, in the full-scale Utube reactor for the ±25%-perturbed kL a values. The volumetric gas flow rates decrease with axial depth, turns up at the column bottom and increases with decreasing axial depth in a concave shape. But they are not changed with the perturbed kL a values and not fully recovered at the exit of the outer tube, indicating that the absorbed oxygen amount is 18.5% of their supplied amount. This suggests that the recycle use of the excess oxygen, which is otherwise disposed,
outer tube kLa 0% kLa+25% kLa-25%
kLa-25% [ kLa 0% kLa+25%
0.01
0 0
6367
Commercial plant 0 0
10
20 27.2 20 Axial depth [m]
10
0
Fig. 8. Axial distribution of volumetric gas velocity for a full-scale U-tube reactor.
would improve the cost effectiveness and energy efficiency of the system, thus reducing the environmental impact. 5.3. Ozone absorption efficiency and removal efficiency of odorous material Key factors indicating the reactor performances and operating conditions for ozonation treatment are defined as follows: Ozone absorption efficiency (dimensionless) = (Influent ozone conc.-Effluent ozone conc.)/ Influent ozone conc. Removal efficiency of odorous material (dimensionless) = (Influent odorous material conc. - Effluent odorous material conc.)/ Influent odorous material conc. Ozone dose (kg m−3 )=Influent ozone conc.×Gas/Liquid ratio, Gas/Liquid ratio (dimensionless) = Volumetric gas flow rate/Volumetric liquid flow rate. In order to verify the practical applicability of the scaleup simulation method presently developed, the measured data on the ozone absorption efficiency and the removal efficiency of odorous materials for a pilot and a full-scale U-tube reactors are compared with the calculated ones in the following. In Fig. 9 the measured values of ozone absorption efficiency for the uses of ozonated air and ozonated oxygen in the pilot plant reactor are correspondingly compared with the calculated ones. Note that the water temperature was considered to evaluate the mass transfer and equilibrium properties for the absorptions of gaseous components in the simulation calculation. It is apparent in the upper figure that the values of ozone absorption efficiency for the ozonated oxygen system are much higher than those for the ozonated air system as shown in the pilot plant data. Also it is shown that the coefficient of variation (CV) value indicating the degree of relative deviation between the measured and calculated values of the ozone absorption efficiency is only 1.24% for the treatment
6368
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
100
100 Removal efficiency of odorous material (cal.) [%]
Ozone absorption efficiency (cal.) [%]
Pilot plant 90
80
70
60 Ozonated air (CV= 3.91%) Ozonated oxygen (CV= 1.24%)
80
60
40
20 CV=16.96 [%] 0
50 50
60
70
80
90
100
0
20 40 60 80 Removal efficiency of odorous material (obs.) [%]
100
100 Fig. 10. Comparison between the observed values of removal efficiency of odorous material 2-MIB and calculated ones for a pilot U-tube reactor.
Ozone absorption efficiency (cal.) [%]
Commercial plant
99
5.4. Correlations of the reactor performance indexes of the U-tube 98
97 Ozonated oxygen (CV=0.32%) 96 96
97 98 99 Ozone absorption efficiency (obs.) [%]
100
Fig. 9. Comparison between the observed values of ozone absorption efficiency and calculated ones for pilot and full-scale U-tube reactors.
using ozonated oxygen while that is 3.91% for the treatment using ozonated air. The ozone absorption efficiency data for the full-scale plant are also compared with calculated ones in the lower figure. Note that the values of the ozone absorption efficiency is high (about 99%) and the coefficient of variation for the ozone absorption efficiency data is only 0.32%, showing that the prediction of the ozone absorption efficiency is excellent for the full-scale deep U-tube. In Fig. 10, the experimental values of the removal efficiency of the odorous materials measured in the pilot plant using ozonated oxygen for the treatment were compared with the calculated ones. As a result the CV value indicating the relative deviation between the calculated and observed values for the removal efficiency was accounted as 17%.
Fig. 11 shows the effect of gas/liquid ratio on the ozone absorption efficiency; in the upper figure the calculated results are shown of the pilot plant while in the lower figure shown of the full-scale plant. The ozone absorption efficiency steadily decreases with increasing gas/liquid ratio, while slightly decreasing with increasing influent ozone concentration. In the full-scale plant with a depth of 27.2 m, the ozone absorption efficiency is much higher than that in the pilot plant with a depth of 17 m and the effect of the ozone concentration on the ozone absorption is quite small. This may be due to increased hydrostatic pressure, which enhances the absorption of ozone. It may be noted that the ozone absorption efficiency can be well correlated by a single parameter of gas/liquid ratio respectively for the specified U-tube unit configurations. Fig. 12 shows the diagrams indicating the effect of ozone dose on the removal efficiency of odorous material 2-MIB; the upper figure shows the variation of removal efficiency of odorous material for the pilot plant while the lower figure shows that for the full-scale plant. The removal efficiency steadily increases with increasing ozone dose (loading), sharply in the range of lower ozone dose and gradually approaches the line of 100% in the range of higher ozone dose, while slightly increasing with increasing influent ozone concentration. It may be noted that the odorous material removal efficiency can be well correlated with a single parameter of ozone dose depending on specified U-tube configurations.
Pilot plant Inlet ozone concentration 0.05 0.1 0.15 [kg/m3]
90
80
70 (a)
0
0.025
0.05
100 Ozone absorption efficiency [%]
Commercial plant
90
80 Inlet ozone concentration 0.05 0.1 0.15 [kg/m3] 70
(b)
0
0.025 Gas/liquid ratio [-]
0.05
Fig. 11. Effect of gas/liquid ratio on ozone absorption efficiency for pilot and full-scale U-tube reactors.
6. Concluding remarks A novel simulation model was constructed of the U-tube ozonation reactor for treating drinking water by assuming that in the inner tube, the flows of gas and liquid are both in plug flow modes, while in the outer tube, the gas phase flow is in the plug flow mode and the liquid phase flow is approximated by the tanks in series model. The effects of hydrostatic pressure on the gas-phase volumetric flow rate and on the gas–liquid equilibriums for ozone and other inactive components were taken into consideration at any axial position. The ozone absorption, the reactions between dissolved ozone and dissolved organic species in the liquid phase, and the hydrodynamics and fluid mixing for the flows of gas and liquid phases were combined to construct the multiple differential mass balance equations in the inner section, and multiple difference mass balance equations for the tanks in series in the outer column section. Available data and correlations in the literatures on the reaction kinetics, gas–liquid equilibriums and hydrodynamic and mass transfer properties are incorporated. The unavailable data on the gas–liquid mass transfer and fluid mixing properties in the inner tube
Removal efficiency of odorous material [%]
Ozone absorption efficiency [%]
100
100
Removal efficiency of odorous material [%]
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
100
6369
(a) Pilot plant
90
80
70 Inlet ozone concentration 0.05 [kg/m3]
60
0.1
0.15
50 0
0.002
0.004
0.006
0.008
(b) Commercial plant 90
80
70 Inlet ozone concentration [kg/m3]
60
0.05
0.1
0.15
50 0
0.002
0.004
0.006
0.008
Ozone dose [kg/m3] Fig. 12. Effect of ozone loading on removal efficiency of odorous material 2-MIB for pilot and full-scale U-tube reactors.
and in the outer column sections are newly experimentally obtained and incorporated. The simulation results well explained the available data of the ozone absorption efficiency and the removal efficiency of the odorous material in a pilot U-tube reactor. The simulation procedure was also successfully extended to verify the performance of a full-scale U-tube reactor for the ozone absorption efficiency. It was shown that the ozone absorption efficiency can be correlated as a single function of the gas/liquid ratio and the removal efficiency of the odorous material can be correlated as a single function of the ozone dose, respectively, in a practical level depending on specified deep U-tube configurations.
Notation CG CL CL∗ dvs D DL
gas phase concentration, kg m−3 liquid phase concentration, kg m−3 equilibrium concentration, kg m−3 sauter diameter, m diameter, m molecular diffusivity, m2 s−1
6370
Dz EL g H G J kL a kd ko kr ks L m M N Pe P P0 Pt R tp T U X z
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
axial dispersion coefficient, m2 s−1 energy dissipation rate per unit mass of liquid, m2 s−3 gravitational acceleration, m s−2 Henry’s constant, Pa mol frac.−1 molar flow rate, mol m−2 s−1 number of mixing tanks, dimensionless volumetric mass transfer coefficient, s−1 decomposition rate const. for ozone consumption substances, m3 kg−1 s−1 decomposition rate const. for 2-MIB, m3 kg−1 s−1 decomposition rate const. for dissolved organic substances, m3 kg−1 s−1 self-decomposion rate constant for the liquid, m1.5 kg−0.5 s−1 column height, m Henry’s constant, dimensionless molecular weight, kg mol−1 mass transfer flux, m3 kg−1 s−1 Peclet number, dimensionless pressure, Pa atmospheric pressure, Pa total pressure, Pa gas constant, Pa m3 mol−1 K −1 average residence time, s absolute temperature, K superficial gas velocity, m s−1 dissolved organic substance concentration, kg m3 axial height, m
Greek letters
G p w
gas holdup, dimensionless viscosity, Pa s density, kg m3 surface tension, kg s−2 standard deviation of R.T.D., s wall shear rate, kg m−1 s−2
Subscripts e G i int j L O out
exit gas ith tank inner tube gaseous species liquid odorous material outer column
S 0
ozone consuming substances inlet,standard condition
Acknowledgements The present work was partly supported by a Grant-in-Aid for Scientific Research (C) (No. 17510077) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The authors are grateful to Hanshin Suido Kigyoudan for providing the reactor performance data of the fullscale U-tube treating drinking water. References Akita, K., Yoshida, F., 1973. Gas holdup and volumetric mass transfer coefficient in bubble columns. Industrial Engineering Chemistry Process Design Devices 12, 76–80. Deckwer, W.-D., 1992. Bubble Column Reactors. Wiley, New York. Miyahara, T., Hirokawa, M., Ueda, M., Yoshida, H., 1994. Solubility of ozone into water in a bubble column. Kagaku Kogaku Ronbunshu 20, 497–503. Moniwa, T., Okada, M., Motoyama, N., Morioka, M., Hohsikawa, H., 1991. Preprint of 42nd Meeting of Japan Waterwarks Association, vol. 4(18). pp. 145–147. Morioka, M., Hoshikawa, H., Motoyama, N., Okada, M., Moniwa, T., 1991. Ozone absorption model of a cross flow contacting vessel. Suidou Kyoukai Zasshi 60 (7), 7–17. Morioka, T., Motoyama, N., Hoshikawa, H., Murakami, A., Okada, M., Moniwa, T., 1993. Kinetic analysis on the effects of dissolved inorganic and organic substances in raw water on the ozonation of Geosmin and 2-MIB. Ozone Science & Engineering 15, 1–18. Morooka, S., Ikezumi, K., Kato, Y., 1978. The decomposition of ozone in aqueous solution. Kagaku Kogaku Ronbunshu 4, 377–380. Muroyama, K., Norieda, T., Morioka, A., Tsuji, T., 1999. Hydrodynamics and computer simulation of an ozone oxidation reactor for treating drinking water. Chemical Engineering Science 54, 5285–5292. Roustan, M., Line, A., Brodard, E., Duguet, J.P., Mallevialle, J., 1990. Theoretical approach and experimental results obtained for a new ozonation gas liquid reactor: the deep U-tube. Water Supply 8 (3–4, Water Nagoya’ 89), 458–464. Roustan, M., Line, A., Duguet, J.P., Mallevialle, J., Wable, O., 1992a. Practical design of a new ozone contactor: the deep U-tube. Ozone Science & Engineering 14, 427–438. Roustan, M., Line, A., Wable, O., 1992b. Modeling of vertical downward gas–liquid flow for the design of a new contactor. Chemical Engineering Science 47 (13/14), 3681–3688. Roustan, M., Beck, C., Walbe, O., Duguet, J.P., Mallevialle, J., 1993. Modeling hydraulics of ozone contactors. Ozone Science & Engineering 15, 213–226. Sato, A., 1992. Water treatment-its new development. Gihoudou Shuppan, (in Japanese). Society of Chemical Engineers Japan (Eds.), 1999. Kagaku-Kogaku Benran, second ed. Somiya, I., 1989. Water treatment using ozone. Kogai Taisaku Gijutu Doyukai, (in Japanese). Wilke, C.R., Chang, P., 1955. Correlation of diffusion coefficients in dilute solutions. A.I.Ch.E. Journal 1 (2), 262–270.
Modeling and scale-up simulation of U-tube ozone oxidation reactor for treating drinking water Katsuhiko Muroyamaa,∗ , Masahiro Yamasakia , Mamoru Shimizua , Eiji Shibutania , Takeshi Tsujib a Department of Chemical Engineering, Kansai University, Osaka 564-8680, Japan b Aqua Technology Research Dept., JFE Engineering Corporation, Kawasaki City, Kanagawa 210-0855, Japan
Received 1 November 2004; received in revised form 20 April 2005; accepted 20 April 2005 Available online 20 June 2005
Abstract In the present study, we developed a novel simulation model of the U-tube reactor for treating drinking water, which is composed of a coaxial inner tube serving as an efficient concurrent down-flow ozone dissolver and an outer column carrying out reactions between ozone and organic substances including odorous materials (2-methylisoborneol: 2-MIB) dissolved in the raw water. We assume that the U-tube is composed of a plug flow section (inner tube) followed by a tanks-in-series section (outer bubble column) and take into account the effect of the hydrostatic pressurization on the flow and absorption equilibrium for the gaseous components including ozone and other inactive species in developing the mass balance models. An algorithm is constructed of the differential multiple mass balance equations for the inner tube sections and multiple difference mass balance equations in the series tanks in the outer column section to enable the scale-up from a pilot plant to a full-scale plant. The gas holdup and gas–liquid mass transfer coefficient were measured in a model reactor and correlated for the use of the simulation calculation. Available literature data and correlations on the rates of reactions between ozone and organic substances including odorous material 2-MIB, gas–liquid equilibrium for active and inactive gases and axial fluid mixing properties are also incorporated in the simulation calculation. The simulation results well explained the available data of the ozone absorption efficiency and the removal efficiency of the odorous material in a pilot U-tube reactor. The simulation procedure was also successfully extended to verify the performance of a full-scale U-tube reactor. It is shown that the ozone absorption is practically a single function of the gas/liquid ratio while the removal efficiency of the odorous material is a single function of the ozone dose for a specified U-tube configuration. 䉷 2005 Elsevier Ltd. All rights reserved. Keywords: Drinking water; Ozone treatment; U-tube reactor; Simulation; Scale-up; Volumetric gas–liquid mass transfer coefficient; Gas holdup
1. Introduction Due to utrophication the increased inclusion of nitric compounds and humic substances occurs in the river water, inevitably resulting in an excess chlorination which in turn generates unwanted chlorine compounds such as trihalomethanes in the treated water. In an advanced drinking water treatment, the employment of an ozone treatment followed by an activated carbon adsorption is becoming popular in Japan (Somiya, 1989). The ozone treatment can reduce the odorous materials such as geosmin and 2-methylisoborneol (2-MIB) and the precursors to the ∗ Corresponding author. Tel.: +81 6 6368 0945; fax: +81 6 6388 8869.
E-mail address: [email protected] (K. Muroyama). 0009-2509/$ - see front matter 䉷 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2005.04.064
trihalomethanes, and compensate for the chlorination treatment, thus reducing the occurrence of the total chlorinated organic compounds. Note that ozone is an environmental risk free substance because the dissolved ozone in water readily self-decomposes and reduces to oxygen (Sato, 1992). U-tube ozonation reactor is a novel bubble column constructed of a concentric downflow tube and a outer column; in the downflow tube ozonated air or ozonated oxygen is mixed with fresh river water and an efficient dissolution of ozone is achieved under hydrostatical pressurization, while in the outer column the dissolved ozone reacts with the precursors to trihalomethanes and the odorous materials such as geosmin and 2-MIB for a sufficient residence time to reduce the concentrations of these target materials within
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
an allowable level. Hydrodynamics of the U-tube reactor was studied on the flow pattern and the gas holdup in the concurrent downflow tube was measured and correlated by Roustan et al. (1990, 1992a,b). The liquid mixing properties in a full scale U-tube reactor were also studied by Roustan et al. (1993). In our previous study (Muroyama et al., 1999), we proposed the design model of a U-tube ozone oxidation reactor for treating drinking water, assuming that in the inner tube, the flows of gas and liquid are both in plug flow mode and in the outer tube, the flow of the gas is still in plug flow mode while the flow of the liquid can be modeled by tanks in series. The proposed model well simulated a pilot U-tube reactor, verifying that the calculated results well predicted the data on the ozone absorption efficiency and the removal efficiency of odorous material 2-MIB. In the present study, we developed a novel design model for the U-tube reactor considering the effect of hydrostatical pressurization on the volumetric flow and the gas–liquid equilibriums for gaseous components including ozone and other inactive species. The ozone absorption, the reaction between dissolved ozone and dissolved odorous material 2MIB in the liquid phase, the flow and mixing of both gas and liquid phases are combined to set up the multiple differential mass balance equations in the inner tube and the multiple mass balance difference equations based on the tanks in series model in the outer column. The physical absorption of inactive gases is also considered in the mass balance equations. An algorithm is constructed of the multiple mass balance equations with appropriate boundary conditions in the two reactor sections and solved to simulate the reactor performance characteristics including the ozone absorption efficiency and the decomposition efficiency of the odorous material for the U-tube ozonation reactor treating drinking water. Available data on the reaction kinetics, gas–liquid equilibrium, absorption rates of gases and fluid mixing reported in the literature or experimentally obtained by ourselves are also incorporated in the simulation calculations. The reactor performance characteristics evaluated for the pilot plant are well predicted by the simulation calculations. The simulation model is also successfully extended to verify the reactor performance for a full-scale U-tube reactor treating drinking water.
2. Experimental 2.1. Experimental apparatus Fig. 1 shows the experimental U-tube apparatus for measuring the hydrodynamic properties and the gas–liquid mass transfer characteristics. The dimensions of the experimental column are as follows; the outer column is 454 mm in I.D. and 3550 mm in height and the inner tube is 75 mm in I.D. and its end is opened at 100 mm above the outer column bottom. The inner tube is fitted coaxially in the outer col-
6361
1 4
6
5
2 3
1 Main column 2 liquid reservoir
4 Liquid flow meter 5 Air supply
3 Liquid pump
6 Gas flow meter
Fig. 1. Experimental apparatus for measuring hydrodynamic and mass transfer characteristics.
Table 1 Operating conditions for experimental system for measuring hydrodynamics and gas–liquid mass transfer characteristics Inner tube Outer column
Superficial Superficial Superficial Superficial
liquid velocity (m/s) gas velocity (m/s) liquid velocity (m/s) gas velocity (m/s)
0.565–1.885 0.0113–0.339 0.0212–0.0739 0.00443–0.0133
umn and the length of its straight portion well exceeds the length of the outer column. The experimental conditions for both phase superficial velocities are listed in Table 1. Two types of gas distributor were used; the first one is a single pipe of 6 mm I.D. stainless steel pipe and the second one is a four-way porous-tips nozzle, welded with sintered porous tips with average pore diameter of 40 m. 2.2. Gas holdup measurement The axial distribution of gas holdup in the inner tube was measured using pairs of electro conductivity cells of thin 25 mm × 60 mm square stainless steel plates which were mounted facing each other on the opposite sides of the inner wall surface at seven axial positions. After a steady state condition for the gas–liquid downflow was established in the inner tube, the inter-cell impedance was measured with 2 kHz AC by using a LCR meter (KC-547, KOKUYO Electric). In advance a correlation between the liquid holdup and the ratio of the inter-cell impedance for the single liquid flow to that for the gas–liquid flow or for the liquid–solid fluidization was obtained. The gas holdup was measured also in the outer column using a static pressure gradient method.
6362
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
2.3. Volumetric mass transfer coefficient, kL a, in the inner tube
G [-]
10-1
nozzle1 nozzle2 UL [m/s]
10-2
0.75 1.13 nozzle1:Single pipe
1.51
nozzle2:Porous -tipsnozzle
1.89
10-3 10-2
10-1 UG [m/s]
10-1
Normal Bubble column Akita-Yoshida(1973)
10-2
10-3 nozzle1:Single pipe nozzle2:Porous -tips nozzle
3. Experimental results for gas holdup and mass transfer
10-4 10-4
10-3
nozzle1 nozzle2 UL [m/s] 0.0296 0.0433 0.0591 0.0793 10-2
10-1
100
UG [m/s]
3.1. Gas holdup The gas holdup, g in the inner tube is plotted against the superficial gas velocity in Fig. 2. The value of g in the concurrent downflow tube increases with superficial gas velocity while it significantly decreases with increasing downward liquid velocity. Note that it is significantly lower than that in the normal bubble column as typically represented by a solid line for Akita–Yoshida’s correlation. This may be explained as follows; in the high shear condition at a high downward liquid velocity the bubbles are finely divided and homogeneously dispersed in the liquid flow. The increase of liquid flow rate steadily increases the downward velocity of gas bubbles and then the gas holdup decreases with increasing liquid velocity. Note that the values of gas holdup for the single pipe nozzle are significantly greater than those for the four-way porous-tips nozzle since the former nozzle creates course bubbles with a wide size distribution while the latter nozzle causes fine bubbles with a narrow distribution. A regression analysis leads to the following empirical gas holdup correlations for the two gas distributors: 0.998 −1.115 gi = 0.821Ugi ULi
100
Fig. 2. Variation of G versus UG in the inner tube.
G [-]
Volumetric mass transfer coefficient in the inner tube was measured by physical absorption of pure oxygen; pure oxygen gas was introduced at a controlled rate through the gas nozzle at the top of the inner tube under a controlled liquid flow, while nitrogen gas was bubbled through a ring-shaped sparger fitted at the bottom of the outer column at a sufficiently high rate to desorb oxygen from the water in the outer column. After a steady state was established, the dissolved oxygen concentrations at the inlet and outlet of the inner tube were measured simultaneously by means of DO meters (YAI Model 57, YSI Japan). The axial differential mass balance equations are formulated to solve the oxygen concentrations in the liquid and gas phases based on the assumption that the gas and liquid flow downward in plug flow modes and the value of the mass transfer coefficient kL a is a constant along the entire inner tube length, while the effect of hydrostatic pressure at any axial position on the volumetric gas flow rate and the gas–liquid equilibrium is considered. The multiple first order differential equations were numerically solved to obtain an optimal value of kL a by an iterative procedure by fitting the calculated outlet dissolved oxygen concentration with the observed one. The mass balance equations and the iterative procedures to solve their equations are the same as those described for the absorption of inactive gases in the later section and omitted here.
Normal Bubble column Akita-Yoshida(1973)
(single-pipe nozzle),
(1)
Fig. 3. Variation of G versus UG in the outer column. 1.066 −1.153 gi = 0.731Ugi ULi
(porous-tips nozzle).
(2)
Fig. 3 shows the plot of the gas holdup values in the outer column against the superficial gas velocity and compares the calculated value from the correlation by Akita and Yoshida (1973) for the normal bubble column. It is obvious that the experimental values are fairly well explained by the Akita–Yoshida’s correlation; however, the experimental gas holdup values are laying slightly below except in a range of lower gas velocities less than 3 × 10−3 m/s, where the experimental gas holdup values slightly exceed the calculated those from Akita–Yoshida’s equation. 3.2. Mass transfer characteristics in the inner tube Fig. 4 shows the plots of kL a value versus UG in the inner tube for two types of gas distributors indicating that the values of kL a increases with UG . The values of kL a in the inner tube are significantly greater than those in the nor-
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
Differential mass balance for the gas phase ozone is represented as dGO3 ∗ − CLO3 ). (6) = −(kL a)O3 (CLO MO3 3 dz
kLa [1/s]
10-1 Normal Bubble column Akita-Yoshida(1973)
10-2
nozzle1: Single pipe nozzle2: Porous-tips nozzle
1.13 1.51
UL
1.89 10-2
10-1
CGO3 = GO3 MO3 /UG .
(7)
Mass balance equation for the liquid-phase inactive gas component (j; oxygen) is given by
0.75
10-3
Here, the liquid phase and gas phase ozone concentrations, ∗ CLO and CGO3 are defined as 3 ∗ = mO3 CGO3 , CLO 3
nozzle1 nozzle2 UL [m/s]
10-4
6363
100
UG [m/s]
dCLj ∗ = (kL a)j (CLj − CLj ). dz
Overall mass balance for the specified gaseous inactive component (j; oxygen) is given by Gj = Gj 0 − UL (CLj − CLj 0 )/Mj .
Fig. 4. Variation of kL a versus UG .
(8)
(9)
The saturated liquid phase concentration of the inactive gas, ∗ , is defined as CLj mal bubble column as typically represented by a curve of Akita–Yoshida’s correlation. The values of kL a are slightly greater for the four-way porous-tips nozzle than for the single pipe nozzle; this is because the bubbles evolved from the four-way porous-tips nozzle are smaller and uniform in size distribution resulting in the higher specific interfacial area. The empirical correlations were formulated for the two types of nozzles as follows: 0.808 Single nozzle: kL a = 0.563UL0.121 UG
Porous-tips nozzle: kL a
0.788 = 0.600UL−0.242 UG
(3) (4)
∗ = mj CGj , CLj
CGj = Gj Mj /UG .
(10)
The total molar flow rate is given by Gt = GO3 + Gj .
(11)
The mol fraction and partial pressure of gaseous species are defined as Gj yj = , pj = Pt yj . (12) Gt Static pressure, Pt (Pa) at any depth is represented as Pt = P0 + L (1 − G )gz.
(13)
Superficial gas velocity is defined as 4. Simulation and scale-up modeling of U-tube reactor Considering the differences in the liquid mixing behaviors between the inner tube section and the outer column section, we employed the plug flow model for the liquid flow in the inner tube section and the series in tanks model in the outer column section, while the gas phase remains to be in a plug flow state in both sections. Mass balance modeling and necessary parameter setting in the two reactor sections will be described for the particular U-tube operation with ozonated oxygen in the following.
UG = Gt RT /Pt
(14)
Boundary conditions at the inner tube inlet are given for the gaseous species as follows: z = 0,
GO3 = GO3 0 ,
Gj = Gj 0 ,
CGO3 = CGO3 0 ,
CLj = CLj 0 .
(15)
Differential mass balance for the ozone consuming substance is given by UL
dXs = −(1 − G )kd CLO3 Xs . dz
(16)
4.1. Mass balances in the inner tube section
Differential mass balance for the odorous material 2-MIB is given by
Differential mass balance for the liquid phase ozone is formulated as
UL
UL
dCLO3 ∗ − CLO3 ) − (1 − G ) = (kL a)O3 (CLO 3 dz 1.5 + kr CLO3 Xs ). × (ks CLO 3
dXO = −(1 − G )kO CLO3 XO . dz
(17)
Boundary conditions for the ozone consuming substance and odorous material are expressed as (5)
z = 0,
Xs = Xs0 ,
XO = XO0 .
(18)
6364
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
4.2. Mass balances in the outer column
Superficial gas velocity is defined by
Mass balances for gas phase and liquid phase in the ith continuous tank are described below. Mass balance for the liquid phase ozone is expressed as UL (CLO3 ,i − CLO3 ,i−1 ) = MO3 (GO3 ,i−1 − GO3 ,i ) 1.5 + kr CLO3 ,i XS,i ). − (1 − G )Li (ks CLO 3 ,i
dGO3 ∗ = −(kL a)O3 (CLO − CLO3 ,i ). 3 dz
(19)
(20)
∗ , Here, the liquid phase saturated ozone concentration, CLO 3 is given by the following relations: ∗ CLO 3
= mO3 CGO3 ,
CGO3 = GO3 MO3 /UG .
(30)
Material balance for ozone consuming substance is given by UL (XL,i−1 − XL,i ) = (1 − g )Li kd CLO XS,i .
(31)
Mass balance for odorous material 2-MIB is given by
Differential mass balance for the gas phase ozone is presented as MO3
UG = Gt RT /Pt .
(21)
UL (XO,i−1 − XO,i ) = (1 − g )Li ko CLO XO,i
(32)
from which, the concentration of odorous materials can be readily determined. 4.3. Gas–liquid equilibrium and mass transfer characteristics The dimensionless Henry’s law constant, mj [-] is defined as follows: RT L [Pa m3 /(mol K)][K][kg/m3 ] = Hj MH2 O [Pa/mol frac.][kg/mol] = [mol frac.],
Eq. (20) can be integrated with the boundary condition at the inlet of the tank as given by
mj =
z = 0, GO3 = GO3 ,i−1 .
where Henry’s law constants for the inactive gases Hj (Pa) was estimated from the correlations given in a handbook (Society of Chemical Engineering Japan, 1999). That for ozone-water system associated with self-decomposition of ozone was estimated from the correlation given by Miyahara et al. (1994). The volumetric mass transfer coefficients for the gaseous species except oxygen in the inner tube were estimated from the correlation obtained from the physical oxygen absorption, (kL a)O2 , as follows: (kL a)j DLj 0.5 = , (34) (kL a)O2 DLO2
(22)
Differential mass balances for the inactive gas component (j; oxygen) are expressed as Mj
dGj ∗ − CLj ,i ), = −(kL a)j (CLj dz
(23)
∗ is given by the following relations: where CLj ∗ CLj = mj CGj ,
CGj = Gj Mj /UG ,
(24)
where mj is dimensionless Henry’s law constant for the inactive gas component. Eq. (23) can be integrated with a boundary condition at the bottom of the ith tank as given by z = 0, Gi = Gj,i−1 .
(25)
From the net mass balance, the gas phase molar flow rate at the exit of ith tank is given by Gj,i = Gj,i−1 − UL (CLj ,i − CLj ,i−1 )/Mj .
(26)
Note that we can determine the outlet liquid phase concentrations of ith tank, CLj ,i which is a constant throughout the tank compare the value of Gj,i obtained by integrating Eq. (23) with that from Eq. (26) to determine. The total molar flow rate is given by Gj . (27) Gt = GO3 + The mol fraction and partial pressure of gaseous species are defined as yj =
Gj , Gt
pj = Pt yj .
(28)
Static pressure at any depth is expressed as Pt = P0 + L (1 − G )gz.
(29)
(33)
where the diffusivities of the gaseous species were estimated from the Wilke–Chang correlation (Wilke and Chang, 1955). The volumetric mass transfer coefficient in the outer column was estimated from the correlation by Akita and Yoshida (1973) because the bubble flow mode and the gas holdup behavior in the outer column were the same to those in the normal bubble column because of low liquid velocity. The gas holdup in the inner tube for the single pipe nozzle was calculated from the correlation presented in Section 2.1. The gas holdup in the outer column was estimated by the correlation presented by Akita and Yoshida (1973). 4.4. Mixing parameter in the outer tube The number of series tanks indicating extent of liquid mixing in the outer tube section was evaluated as follows. The axial dispersion coefficient, Dz , in the normal bubble column can be estimated by the correlation of Deckwer (1992) which covers a range of lower superficial gas velocities as exemplified in the outer column of the U-tube. The dimensionless standard deviation for the normalized residence time
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
distribution for the axial dispersion model can be related to the inverse of the number of mixing tanks, J for the series in tanks model as follows: 2P 2 1 2 − (1 − exp(−P e) = , (35) = 2 Pe Pe J tP where Pe is the Peclet number for the axial dispersion model, P e = UL L/Dz . From Eq. (35), a real number is obtained for a specified gas–liquid operating conditions. In the simulation calculation of the tanks in series model, an additional reactor corresponding to the decimals part of the real number, j, was connected to the last one of the equivalent volume reactors corresponding to the integer part of j. 4.5. Kinetic parameters for the ozone oxidation reactions The self decomposition rate can be approximated by the first to second order kinetics with respect to the dissolved ozone concentration in the neutral water as encountered for the drinking water treatment (Morooka et al., 1978; Morioka et al., 1991; Miyahara et al., 1994). It was found that the correlation equation for the kinetic constant proposed by Morioka et al. deviates from other two correlations in the range of higher ozone concentration. Thus we employed the correlation of the kinetic constant for the ozone selfdecomposition most recently presented by Miyahara et al. (1994) whose correlation is shown below: ks = (4.0 × 1011 + 9.6 × 1014 [OH− ]0.8 ) × exp(−9.6 × 103 /T ).
−1 −1
kd = 0.8333 (m kg 3
s
).
Simulation parameter
Pilot plant
Full-scale plant
Column height (m) Inner tube diameter (m) Outer column diameter (m) Gas flow rate (m3 /s) Liquid flow rate (m3 /s) Liquid temperature (◦ C) Ozone concentration (kg/m3 ) Concentration of ozone consuming substance (kg/m3 ) Residence time in inner tube (s) Residence time in outer column (s) Number of mixing tanks in outer column (number)
17.0 0.047 0.312 1.04×10−4 5.36×10−3 20 0.148 1.60×10−3
27.2 0.77 3.95 1.89×10−2 0.944 20 0.148 1.60×10−3
5.5 237 18.6
13.4 340 4.38
which the axial distributions of species involved in the ozone oxidation process, such as, liquid phase ozone concentration, gas phase, and liquid phase concentration of odorous materials are numerically calculated. Note that the operating conditions listed in Table 2 correspond to the case where the ozonated oxygen is used to treat drinking water. The water temperature in the calculation was assumed to be a constant value of 20 ◦ C in the following calculations. 4.7. Algorithm and calculation procedure
(37a) (37b)
First order rate constant for the reaction of ozone with dissolved odorous material 2-MIB was correlated by Morioka et al. (1991) as follows: ko = exp(45.0 + 0.9[pH]) exp(−11100/T ).
Table 2 Reactor dimensions and operating conditions for simulation calculation
(36)
Note that Miyahara et al. determined that the ozone decomposition rate is proportional to 1.5th order with respect to the dissolved ozone concentration. Moniwa et al. (1991) investigated the first order reaction rate constant for the ozone consumption rate constant and the decomposition rate constant for the reaction of ozone with dissolved organic materials as follows: kr = 0.6667 (m3 kg−1 s−1 ),
6365
(38)
Note that Morioka et al. (1993) investigated the kinetics of ozone decomposition for geosmin and 2-MIB under the presence of carbonate, free chlorine, alcohols and volatile fatty acids. 4.6. Operating conditions The major dimensions, and operating conditions for the pilot and full-scale U-tube reactors are listed in Table 2, for
In the inner tube section, the axially differential equations for liquid-phase and gas-phase components including ozone, inactive gaseous components, and dissolved ozone consuming substances were solved by the Runge–Kutta–Gill methods based on the boundary conditions at the inlet of the inner tube for both phases. After determining the axial concentration distribution of these components, the liquidphase concentration of odorous material was solved. In the outer section, the difference mass balance equations for each mixed tank were solved by a trial-and-error procedure to fit the outlet liquid phase concentration with that from the overall mass balance between the phases. After determining the liquid phase concentrations, the concentration of odorous material was determined. The set of calculations were successively carried out to the final tank. The algorithm was constructed based on a C++ -language source program and all the calculations were carried out on a PC.
5. Simulation results In the following, the simulation calculations are carried out for the U-tube reactor operated with ozonated oxygen since the pilot and full-scale plants has been successfully demonstrated with ozonated oxygen.
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
outer column
equilibrium ozone conc. CLO3*=m×CGO3
0.03
Liq.-phase ozone conc. CLO3
kLa-25%
0.02 kLa0%
Pilot plant
0.01 kLa+25%
inner tube 0.003
1
kLa+25%
0.002
kLa0% kLa-25%
10
0
17
Pilot plant
0
0
inner tube
outer column
equilibrium ozone conc. CLO3*=m×CGO3
0.03
Liq.-phase ozone conc. CLO3
kLa-25%
0.02
kLa0%
Commercial plant
0.01 (b)
kLa+25%
0
0
10
20
27.2
20
10
0
10
17
10
[×10-7]
0
Axial depth [m]
(a) Liq.-phase ozone conc. [kg/m3]
0.04 Liq.-phase ozone conc. [kg/m3]
0
Axial depth [m]
(a)
(b)
10
0.5
0.001
0
0
1.5
outer column
Concentration of 2-MIB [kg/m3]
inner tube
Liq-phase ozone conc. [kg/m3]
Liq.-phase ozone conc. [kg/m3]
0.04
inner tube
-7 1.5 [×10 ]
outer column
kLa+25%
0.003
1 kLa0%
0.002
kLa-25%
0.5
0.001 Commercial plant
0
0
10
20 27.2 20 Axial depth [m]
10
0
0
concentration of 2-MIB [kg/m3]
6366
Fig. 6. Variation of dissolved ozone and odorous material 2-MIB for pilot and full-scale U-tube reactors.
Axial depth [m]
Fig. 5. Variation of dissolved ozone concentration with axial height for pilot and full-scale U-tube reactors.
5.1. Axial distribution of gaseous and liquid-phase components Fig. 5 shows the variations of the liquid phase and gas phase ozone concentrations versus axial depth in the inner tube and outer column sections. Note that the equilibrium liquid phase ozone concentration which is the value of product of dimensionless equilibrium constant, m, and the gas phase ozone concentration, m × CGO3 , is shown in place of the real gas phase ozone concentration to indicate how the liquid phase ozone concentration approaches their equilibrium value along the axial depth. In the figure kL a-0% means the value of the kL a estimated by the correlation for the single nozzle obtained in this work. Note that the value of kL a + 25% is increased by 25%, and that of kL a − 25% is decreased by 25% from that of kL a − 0%. It is obvious that in both reactors the dissolved ozone concentration increases with increasing axial depth in the inner tube while in the outer column it only slightly decreases with decreasing axial depth. In the outer column of the pilot plant the dissolved ozone concentration approaches their
equilibrium value and eventually exceeds the latter in the top section of the column where the ozone tends to be removed from the liquid into the gas. In the full-scale commercial plant, the dissolved ozone concentration approaches their equilibrium value at an axial depth of about 20 m and afterwards it becomes insensitive to the axial depth in the inner tube, while in the outer column it slightly decreases with decreasing axial depth and eventually exceeds the equilibrium concentration in the upper half region of the column where the desorption of ozone from liquid to gas surely occurs. The increase in kL a value should enhance the dissolution of ozone but the increase in the liquid phase ozone concentration significantly occurs only within an initial 10 m of the axial depth. Note that in the full-scale deep reactor, the influence of ±25% change in kL a on the dissolved ozone concentration is small because of the increased hydrostatic pressurization. Fig. 6 shows the variation of liquid-phase odorous material concentration versus axial depth together with that of liquid-phase ozone concentration. Because of its small outer column diameter, the number of mixing tank in series for the pilot plant is as high as 18.9 and the axial distribution of liquid-phase odorous material concentration is almost constant in the inner tube but it rather continuously decreases with increasing rise of axial depth. On the other hand in the full-scale reactor with a larger diameter of 3.95 m the num-
6
Liquid-phase equilibrium concentration
Commercial plant
5 4
0.1 kLa-25%
3
[ kk a0% a+25% L L
0.05
kLa+25%
2
kLa0%
1 kLa-25%
0
0.02 Volumetric gas velocity [m3/s]
outer column
inner tube
0.15
Gas-phase oxygen conc. [kg/m3]
Liquid-phase oxygen conc. [kg/m3]
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
inner tube
10
20 27.2 20 Axial depth [m]
10
0
Fig. 7. Axial oxygen concentration distributions in the liquid and gas phases for a full-scale U-tube reactor.
ber of the mixing tanks is only 4.4 and the concentration of the liquid-phase odorous material decreases in stepwise while the concentration drop in the first tank being largest. The liquid-phase odorous material concentration in the outer column only slightly varies with the ±25% changes from a kL a value estimated for the specified operating condition. Note that to improve the accuracy in the calculation model for the outer column the series tanks are considered to be composed of not only the number of tanks with an equal volume, corresponding to the integer part of the real number, J, derived from Eq. (35), but also an additional one of a smaller volume equivalent to the decimal fraction of J. 5.2. Axial distribution of inactive gas In Fig. 7, axial distributions of gas-phase and liquidphase concentrations of the inactive gas, oxygen, are shown for the specified conditions in the full-scale U-tube reactor. The gas-phase oxygen concentrations linearly increase with axial depth in the inner tube, and inversely linearly decrease with axial rise in the outer column due to the hydrostatic effect. The axial variations of the gas-phase oxygen concentration are almost neglected even for the ±25%-perturbed kL a values. The values of the liquid phase oxygen concentration also linearly increase with increasing axial depth in the inner tube, while in the outer tube, they remain almost constant values corresponding to the different kL a values. Fig. 8 shows the axial variations of the volumetric flow rate of gas, mainly consisted of oxygen, in the full-scale Utube reactor for the ±25%-perturbed kL a values. The volumetric gas flow rates decrease with axial depth, turns up at the column bottom and increases with decreasing axial depth in a concave shape. But they are not changed with the perturbed kL a values and not fully recovered at the exit of the outer tube, indicating that the absorbed oxygen amount is 18.5% of their supplied amount. This suggests that the recycle use of the excess oxygen, which is otherwise disposed,
outer tube kLa 0% kLa+25% kLa-25%
kLa-25% [ kLa 0% kLa+25%
0.01
0 0
6367
Commercial plant 0 0
10
20 27.2 20 Axial depth [m]
10
0
Fig. 8. Axial distribution of volumetric gas velocity for a full-scale U-tube reactor.
would improve the cost effectiveness and energy efficiency of the system, thus reducing the environmental impact. 5.3. Ozone absorption efficiency and removal efficiency of odorous material Key factors indicating the reactor performances and operating conditions for ozonation treatment are defined as follows: Ozone absorption efficiency (dimensionless) = (Influent ozone conc.-Effluent ozone conc.)/ Influent ozone conc. Removal efficiency of odorous material (dimensionless) = (Influent odorous material conc. - Effluent odorous material conc.)/ Influent odorous material conc. Ozone dose (kg m−3 )=Influent ozone conc.×Gas/Liquid ratio, Gas/Liquid ratio (dimensionless) = Volumetric gas flow rate/Volumetric liquid flow rate. In order to verify the practical applicability of the scaleup simulation method presently developed, the measured data on the ozone absorption efficiency and the removal efficiency of odorous materials for a pilot and a full-scale U-tube reactors are compared with the calculated ones in the following. In Fig. 9 the measured values of ozone absorption efficiency for the uses of ozonated air and ozonated oxygen in the pilot plant reactor are correspondingly compared with the calculated ones. Note that the water temperature was considered to evaluate the mass transfer and equilibrium properties for the absorptions of gaseous components in the simulation calculation. It is apparent in the upper figure that the values of ozone absorption efficiency for the ozonated oxygen system are much higher than those for the ozonated air system as shown in the pilot plant data. Also it is shown that the coefficient of variation (CV) value indicating the degree of relative deviation between the measured and calculated values of the ozone absorption efficiency is only 1.24% for the treatment
6368
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
100
100 Removal efficiency of odorous material (cal.) [%]
Ozone absorption efficiency (cal.) [%]
Pilot plant 90
80
70
60 Ozonated air (CV= 3.91%) Ozonated oxygen (CV= 1.24%)
80
60
40
20 CV=16.96 [%] 0
50 50
60
70
80
90
100
0
20 40 60 80 Removal efficiency of odorous material (obs.) [%]
100
100 Fig. 10. Comparison between the observed values of removal efficiency of odorous material 2-MIB and calculated ones for a pilot U-tube reactor.
Ozone absorption efficiency (cal.) [%]
Commercial plant
99
5.4. Correlations of the reactor performance indexes of the U-tube 98
97 Ozonated oxygen (CV=0.32%) 96 96
97 98 99 Ozone absorption efficiency (obs.) [%]
100
Fig. 9. Comparison between the observed values of ozone absorption efficiency and calculated ones for pilot and full-scale U-tube reactors.
using ozonated oxygen while that is 3.91% for the treatment using ozonated air. The ozone absorption efficiency data for the full-scale plant are also compared with calculated ones in the lower figure. Note that the values of the ozone absorption efficiency is high (about 99%) and the coefficient of variation for the ozone absorption efficiency data is only 0.32%, showing that the prediction of the ozone absorption efficiency is excellent for the full-scale deep U-tube. In Fig. 10, the experimental values of the removal efficiency of the odorous materials measured in the pilot plant using ozonated oxygen for the treatment were compared with the calculated ones. As a result the CV value indicating the relative deviation between the calculated and observed values for the removal efficiency was accounted as 17%.
Fig. 11 shows the effect of gas/liquid ratio on the ozone absorption efficiency; in the upper figure the calculated results are shown of the pilot plant while in the lower figure shown of the full-scale plant. The ozone absorption efficiency steadily decreases with increasing gas/liquid ratio, while slightly decreasing with increasing influent ozone concentration. In the full-scale plant with a depth of 27.2 m, the ozone absorption efficiency is much higher than that in the pilot plant with a depth of 17 m and the effect of the ozone concentration on the ozone absorption is quite small. This may be due to increased hydrostatic pressure, which enhances the absorption of ozone. It may be noted that the ozone absorption efficiency can be well correlated by a single parameter of gas/liquid ratio respectively for the specified U-tube unit configurations. Fig. 12 shows the diagrams indicating the effect of ozone dose on the removal efficiency of odorous material 2-MIB; the upper figure shows the variation of removal efficiency of odorous material for the pilot plant while the lower figure shows that for the full-scale plant. The removal efficiency steadily increases with increasing ozone dose (loading), sharply in the range of lower ozone dose and gradually approaches the line of 100% in the range of higher ozone dose, while slightly increasing with increasing influent ozone concentration. It may be noted that the odorous material removal efficiency can be well correlated with a single parameter of ozone dose depending on specified U-tube configurations.
Pilot plant Inlet ozone concentration 0.05 0.1 0.15 [kg/m3]
90
80
70 (a)
0
0.025
0.05
100 Ozone absorption efficiency [%]
Commercial plant
90
80 Inlet ozone concentration 0.05 0.1 0.15 [kg/m3] 70
(b)
0
0.025 Gas/liquid ratio [-]
0.05
Fig. 11. Effect of gas/liquid ratio on ozone absorption efficiency for pilot and full-scale U-tube reactors.
6. Concluding remarks A novel simulation model was constructed of the U-tube ozonation reactor for treating drinking water by assuming that in the inner tube, the flows of gas and liquid are both in plug flow modes, while in the outer tube, the gas phase flow is in the plug flow mode and the liquid phase flow is approximated by the tanks in series model. The effects of hydrostatic pressure on the gas-phase volumetric flow rate and on the gas–liquid equilibriums for ozone and other inactive components were taken into consideration at any axial position. The ozone absorption, the reactions between dissolved ozone and dissolved organic species in the liquid phase, and the hydrodynamics and fluid mixing for the flows of gas and liquid phases were combined to construct the multiple differential mass balance equations in the inner section, and multiple difference mass balance equations for the tanks in series in the outer column section. Available data and correlations in the literatures on the reaction kinetics, gas–liquid equilibriums and hydrodynamic and mass transfer properties are incorporated. The unavailable data on the gas–liquid mass transfer and fluid mixing properties in the inner tube
Removal efficiency of odorous material [%]
Ozone absorption efficiency [%]
100
100
Removal efficiency of odorous material [%]
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
100
6369
(a) Pilot plant
90
80
70 Inlet ozone concentration 0.05 [kg/m3]
60
0.1
0.15
50 0
0.002
0.004
0.006
0.008
(b) Commercial plant 90
80
70 Inlet ozone concentration [kg/m3]
60
0.05
0.1
0.15
50 0
0.002
0.004
0.006
0.008
Ozone dose [kg/m3] Fig. 12. Effect of ozone loading on removal efficiency of odorous material 2-MIB for pilot and full-scale U-tube reactors.
and in the outer column sections are newly experimentally obtained and incorporated. The simulation results well explained the available data of the ozone absorption efficiency and the removal efficiency of the odorous material in a pilot U-tube reactor. The simulation procedure was also successfully extended to verify the performance of a full-scale U-tube reactor for the ozone absorption efficiency. It was shown that the ozone absorption efficiency can be correlated as a single function of the gas/liquid ratio and the removal efficiency of the odorous material can be correlated as a single function of the ozone dose, respectively, in a practical level depending on specified deep U-tube configurations.
Notation CG CL CL∗ dvs D DL
gas phase concentration, kg m−3 liquid phase concentration, kg m−3 equilibrium concentration, kg m−3 sauter diameter, m diameter, m molecular diffusivity, m2 s−1
6370
Dz EL g H G J kL a kd ko kr ks L m M N Pe P P0 Pt R tp T U X z
K. Muroyama et al. / Chemical Engineering Science 60 (2005) 6360 – 6370
axial dispersion coefficient, m2 s−1 energy dissipation rate per unit mass of liquid, m2 s−3 gravitational acceleration, m s−2 Henry’s constant, Pa mol frac.−1 molar flow rate, mol m−2 s−1 number of mixing tanks, dimensionless volumetric mass transfer coefficient, s−1 decomposition rate const. for ozone consumption substances, m3 kg−1 s−1 decomposition rate const. for 2-MIB, m3 kg−1 s−1 decomposition rate const. for dissolved organic substances, m3 kg−1 s−1 self-decomposion rate constant for the liquid, m1.5 kg−0.5 s−1 column height, m Henry’s constant, dimensionless molecular weight, kg mol−1 mass transfer flux, m3 kg−1 s−1 Peclet number, dimensionless pressure, Pa atmospheric pressure, Pa total pressure, Pa gas constant, Pa m3 mol−1 K −1 average residence time, s absolute temperature, K superficial gas velocity, m s−1 dissolved organic substance concentration, kg m3 axial height, m
Greek letters
G p w
gas holdup, dimensionless viscosity, Pa s density, kg m3 surface tension, kg s−2 standard deviation of R.T.D., s wall shear rate, kg m−1 s−2
Subscripts e G i int j L O out
exit gas ith tank inner tube gaseous species liquid odorous material outer column
S 0
ozone consuming substances inlet,standard condition
Acknowledgements The present work was partly supported by a Grant-in-Aid for Scientific Research (C) (No. 17510077) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The authors are grateful to Hanshin Suido Kigyoudan for providing the reactor performance data of the fullscale U-tube treating drinking water. References Akita, K., Yoshida, F., 1973. Gas holdup and volumetric mass transfer coefficient in bubble columns. Industrial Engineering Chemistry Process Design Devices 12, 76–80. Deckwer, W.-D., 1992. Bubble Column Reactors. Wiley, New York. Miyahara, T., Hirokawa, M., Ueda, M., Yoshida, H., 1994. Solubility of ozone into water in a bubble column. Kagaku Kogaku Ronbunshu 20, 497–503. Moniwa, T., Okada, M., Motoyama, N., Morioka, M., Hohsikawa, H., 1991. Preprint of 42nd Meeting of Japan Waterwarks Association, vol. 4(18). pp. 145–147. Morioka, M., Hoshikawa, H., Motoyama, N., Okada, M., Moniwa, T., 1991. Ozone absorption model of a cross flow contacting vessel. Suidou Kyoukai Zasshi 60 (7), 7–17. Morioka, T., Motoyama, N., Hoshikawa, H., Murakami, A., Okada, M., Moniwa, T., 1993. Kinetic analysis on the effects of dissolved inorganic and organic substances in raw water on the ozonation of Geosmin and 2-MIB. Ozone Science & Engineering 15, 1–18. Morooka, S., Ikezumi, K., Kato, Y., 1978. The decomposition of ozone in aqueous solution. Kagaku Kogaku Ronbunshu 4, 377–380. Muroyama, K., Norieda, T., Morioka, A., Tsuji, T., 1999. Hydrodynamics and computer simulation of an ozone oxidation reactor for treating drinking water. Chemical Engineering Science 54, 5285–5292. Roustan, M., Line, A., Brodard, E., Duguet, J.P., Mallevialle, J., 1990. Theoretical approach and experimental results obtained for a new ozonation gas liquid reactor: the deep U-tube. Water Supply 8 (3–4, Water Nagoya’ 89), 458–464. Roustan, M., Line, A., Duguet, J.P., Mallevialle, J., Wable, O., 1992a. Practical design of a new ozone contactor: the deep U-tube. Ozone Science & Engineering 14, 427–438. Roustan, M., Line, A., Wable, O., 1992b. Modeling of vertical downward gas–liquid flow for the design of a new contactor. Chemical Engineering Science 47 (13/14), 3681–3688. Roustan, M., Beck, C., Walbe, O., Duguet, J.P., Mallevialle, J., 1993. Modeling hydraulics of ozone contactors. Ozone Science & Engineering 15, 213–226. Sato, A., 1992. Water treatment-its new development. Gihoudou Shuppan, (in Japanese). Society of Chemical Engineers Japan (Eds.), 1999. Kagaku-Kogaku Benran, second ed. Somiya, I., 1989. Water treatment using ozone. Kogai Taisaku Gijutu Doyukai, (in Japanese). Wilke, C.R., Chang, P., 1955. Correlation of diffusion coefficients in dilute solutions. A.I.Ch.E. Journal 1 (2), 262–270.