Jinjin Zhang ,Zhengming Gao ,Yating Cai,Ziqi Cai,Jie Yang ,*,Yuyun Bao ,*

1 State Key Laboratory of Chemical Resource Engineering,School of Chemical Engineering,Beijing University of Chemical Technology,Beijing 100029,China

2 School of Chemical Engineering,Shandong University of Technology,Zibo 255000,China

1.Introduction

Mechanically agitated gas–liquid reactors with single or multiple impellers are often used to intensify the contact between gas and liquid.Hydrogenation,chlorination,polymerization,sewage treatment,and aerobic fermentation are examples of such processes.Since the mass transfer of gas into liquid is often the rate-limiting step when low solubility gases are involved,the volumetric mass transfer coefficient,kLa,is considered a key parameter for operation,design,and scale-up of stirred reactors.

Based on the flow patterns generated,an impeller can be characterized as an axial flow impeller such as marine propeller,Techmix 335(TX)[1],Lightnin A315(LTN)[2],four-wide-blade hydrofoil impeller(WH)[3],KHX[4]and CBY[5],and a radial flow impeller such as Rushton turbine(RT),pitched blade(PB),concave-bladed disk turbine(CD)[6],half elliptical-blade disk turbine(HEDT)[7],parabolic blade disk turbine(PDT)[5]and Narcissus(NS)[2].Compared with the stirred reactors equipped with a single impeller,the dual-impeller[8–10]and triple-impeller[11–14]stirred reactors have such advantages as increased gas hold-up,superior gas distribution,long residence of gas bubbles,improved liquid circulation characteristics,low power consumption per impeller,and even distribution of shear stress and energy dissipation,thus,effective gas utilization.Over the past few decades,many publications presenting various hydraulic and mass transfer characteristics in multi-impeller stirred vessels can be found in the literature.Nocentini et al.[11]and Linek et al.[15]used multiple Rushton turbines on a common shaft.Arjunwadkar et al.[8]used dual impeller of RT and PB,and Suhaili et al.[16]used dual impeller of RT and CD.kLa for various impeller combinations was discussed by Moucha et al.[1]using 18 combinations of RT,PB and TX and their combinations,and Fujasová et al.[2]using 28 combinations of RT,PB,TX,LNT and NS and their combinations.Results show that radial flow impellers exhibit20%to 50%higher oxygen transfer efficiency than axial flow impellers.However,radial flow impellers have the weakness of bad homogenization[2,12],and relatively high power consumption[17].Compared with radial flow impeller combinations or axial flow impeller combinations,the mixed flow impeller combinations with both radial and axial flow impellers showed better homogenization performance and mass transfer performance at the same power consumption.However,few reports present the mass transfer characteristics for mixed flow impeller combinations.

In the mixed flow impeller combinations,the radial flow impeller is often installed in the bottom to break and disperse the gas introduced from gas sparger and the axial flow impellers are installed above to circulate the gas–liquid flow.RTis the radial flow impeller usually used[1,2,12],but it has weakness of significant drop in gassed power draw.This disadvantage can be tackled by retro fitting of RT with streamlined impellers,such as impellers of CD,HEDT and PDT.However,research on the mass transfer characteristics for these modified impellers is quite rare[12],especially for impellers HEDT and PDT.Moreover,the superficial gas velocity(uG)for kLa determination reported in multi-impeller stirred reactors is usually small(less than 0.016 m·s-1)and the research on kLa at higher uG(more than 0.016 m·s-1)is open for further investigation.

To date,data on kLa for the mixed flow impeller combinations are still not enough,especially for the new-type impeller combinations and at high gas rates.The aim of this work is to provide gassed power demand and mass transfer characteristics for five mixed flow impeller combinations with new-type axial flow impellers of WH(pumping down and pumping up)and CBY(narrow blade and wide blade)with radial flow impellers of HEDT and PDT at uGbetween 0.0078 and 0.039 m·s-1.General correlations representing the dependence of gassed power and kLa on operating conditions are developed.kLa differences among these mixed flow impeller combinations are discussed and analyzed.

2.Experimental

2.1.Experimental setup

All experiments were carried out in a dished-base cylindrical vessel equipped with triple impellers,as shown in Fig.1.The diameter(T)of the tank was 0.30 m and the height of the tank(HT)was 0.75 m.The ungassed liquid level was maintained at a height H=1.8 T with a deionized water volume of 0.036 m3.As a standard configuration,four 0.03 m-wide baffles were symmetrically mounted in the tank.The impellers used were six-half-elliptical-blade disk turbine(HEDT),parabolic-blade disk turbine(PDT),four-wide-blade hydrofoil impeller(WH),and CBY in Fig.2.The diameter of all the impellers(D)was 0.4 T,the same as the clearance between the lowest impeller and the base of the tank,and the distance between two adjacent impellers was 0.48 T.WH can be used for up-pumping and down-pumping operating modes,identified as WHUand WHD,respectively.CBY can be classified as CBYNand CBYWby the impeller tip width of 0.1D and 0.14D,respectively.The shorthand notation used for defining the agitation combinations is straightforward:PDT+2CBYNmeans parabolic-blade disk turbine at the bottom and two CBY impellers with an impeller tip width of 0.1D at mid-level and top-level.PDT+2CBYW,PDT+2WHD,HEDT+2WHDand HEDT+2WHUrepresent similar implications.Air was introduced from a sparger located at a height of 0.33 T above the tank bottom.The ring sparger was 0.8D in diameter with 27 symmetrical downward-directed holes of 0.002 m diameter.

Fig.1.Experimental setup:(1)1st bottle of Na2SO3 solution;(2)2nd bottle of Na2SO3 solution;(3)peristaltic pump;(4)motor;(5)rotary torque sensor;(6)sparger;(7)DO probe;and(8)computer.

Fig.2.Impellers used in the experiment:(a)HEDT;(b)PDT;(c)WH;and(d)CBY.

The first bottle of Na2SO3solution with a concentration of 500 mol·m-3was used to absorb the oxygen in the air so that no variation of Na2SO3concentration in the second bottle would be resulted.The Na2SO3solution with a concentration of 200 mol·m-3in the second bottle was fed into the tank by a peristaltic pump,and the ion concentration in the stirred tank was kept below 50 mol-ion·m-3to maintain the water as a coalescent system[18].The concentration of dissolved oxygen was measured online by a DO probe(VISIFERM DO Arc 120,Hamilton,Switzerland)and recorded with time serial as data files by the computer.

2.2.Power consumption

The power consumption of the impeller was measured by using a rotary torque sensor(AKC-205,China Academy of Aerospace Aerodynamics,China)installed between the impeller shaft and the motor bearings.The specific power consumption PTm(in W·kg-1),which is the sum of the potential energy of the sparged gas and the agitation power consumption calculated by measuring the torque of the stirring shaft and the agitation speed,is an important parameter to characterize the performance of an agitator.The total power consumption of the stirring system was calculated by

where Pgis the gassed agitation power given by Pg=2πNM and Peis the potential energy of sparged gas calculated by Pe=ρLgHSQg.

2.3.Measurement of kLa

kLa was measured by the steady-state sulfite feeding method(SFM)[18],which is based on the measurement of dissolved oxygen concentrations under steady-state conditions of equilibrium between sulfite addition and oxygen dissolution.kLa is calculated by

where Qsis the feed rate of the sulfite solution,m3·s-1;Csis the concentration of sulfite in the feed solution,mol·m-3;VLis the volume of the liquid in the aerated stirred tank,m3;CAiis the equilibrium concentration of the oxygen at the gas–liquid interface,mol·m-3;and CAsis the oxygen concentration measured in water in steady state with the sulfite solution fed continuously,mol·m-3.The experimental results are reproducible twice or thrice with an average relative error less than 5%.

3.Results and Discussion

The superficial gas velocity in this work ranges from 0.0039 to 0.039 m·s-1(equivalentto 0.463–4.63 vvm).We labeled the superficial gas velocity of 0.0039 to 0.0078 m·s-1as low gas velocity,and 0.024 to 0.039 m·s-1as high gas velocity based on the criterion for industrial operation.The impeller speeds for each impeller combination were above the complete gas dispersion speed(NCD)defined by Nienow et al.[19]during all of the measurements of power demand and volumetric mass transfer coefficient.The highest NCDat the highest super ficial gas velocity of 0.039 m·s-1for all impeller combinations used is 6.83 s-1,and the impeller speed used in the experiments was chosen from 8 to 13 s-1.

3.1.Relative power demand(RPD)

The ratio of gassed to ungassed power(RPD,RPD=Pg/P0)for different impeller combinations at various agitation speeds is shown in Fig.3.It can be seen that RPD decreases with increasing gas flow number(FlG)for all impeller combinations because as more gas is introduced into the stirred tank,the gas holdup increases and the average density of the gas–liquid mixture phase decreases.The decreasing average density causes the decrease of RPD with increasing FlG.Under all operating conditions,RPD of all impeller combinations is mostly above 0.65,higher than those of impeller combinations only using the radial impellers or axial impellers reported in the literature[20],meanwhile RPD of axial impellers 3WHDdecreases dramatically to 0.4–0.5 when FlGis larger than 0.05 and RPD of radial impellers 3CD is smaller than 0.55 when FlGis larger than 0.10.The higher RPD obtained in this research shows mixed flow impeller combinations having higher energy utilization efficiency and less loss of capacity for mass transfer.

Power number(NP=P0/ρLN3D5)is one of the most basic parameters for stirred reactors and the gassed power number(NPG=Pg/ρLN3D5)is for the power number under aeration.Based on the experimental data of gassed power under different conditions,we use the following correlation

to regress NPGwith the Froude number and gas flow number for different impeller combinations,and the results are shown in Table 1.

Exponent b indicates the sensitivity of RPD on gas flow for different impeller combinations and is affected by the cavities behind impeller blades or/and the average density of the medium.The effects of gas flow and agitation speed on RPD for different impeller combinations will be discussed from the following three aspects.

3.1.1.Influence of bottom impeller

In Fig.3,RPD for PDT+2WHDand HEDT+2WHDpresents the similar trend for different agitation speeds although with different bottom impellers.Both PDT and HEDT impellers are streamlined modification of RT with different curvatures.Vasconcelos et al.[6]studied the effect of blade shape on the performance of impellers and found that the blade streamlining can lead to a lower ungassed power number(NP).Therefore,it can be referred that NPfor PDT is lower than HEDT.In this work,NPGfor these two impeller combinations of PDT+2WHDand HEDT+2WHDwas regressed in Table 1,while NPfor PDT+2WHDand HEDT+2WHDwithout aeration was also obtained as 2.8 and 3.6,respectively.It can be seen that both NPand NPGfor PDT+2WHDare much lower than HEDT+2WHDat the same operation conditions.This indicated that the bottom impeller can influence greatly the power consumption.Nevertheless,the similar RPD for PDT+2WHDand HEDT+2WHDshows that the influence of the bottom impeller on RPD is little.

3.1.2.Influence of blade width

The effect of the blade width on RPD can be seen by the comparison of RPD for PDT+2CBYN,PDT+2CBYWand PDT+2WHD,having the same bottom impeller and increasing width of blade for two upper impellers.Table 1 shows that both b and c increase with increasing blade width of upper impellers from CBYN,CBYWto WHDwith the same bottom PDT impeller.The increasing values of b and c imply that RPD decreases more when more gas is introduced or stirred at higher agitation speed.It could be explained that the wider blade of upper impellers can supply bigger interaction area for both liquid and gas,introduce more gas into the liquid and keep bubbles staying for a longer time in liquid,thus the gas holdup increases,so RPD decreases more when FlGor Fr increases.

3.1.3.Influence of up-and down-pumping operating modes

The obviously different RPD trends for HEDT+2WHUand HEDT+2WHDshow that the operating mode for the axial upper impellers can influence RPD greatly.RPD for HEDT+2WHUare all above 0.75,much higher than those for other impeller combinations.That is because the flow field of liquid caused by HEDT+2WHU,which was computed by Min et al.[21],has the same direction of the gas flow,leading to a rapid release of bubbles from the liquid,thus to a reduction in the retained gas and further to a high average medium density.The smaller exponent b for HEDT+2WHUthan HEDT+2WHDin Table 1 indicates that the gas flow rate has less influence on RPD for HEDT+2WHUthan HEDT+2WHDbecause of the shorter residence time of retained gas in liquid and lower gas holdup for HEDT+2WHU.From the comparison of the different RPD trends for HEDT+2WHUand HEDT+2WHD,and the similar RPD for PDT+2WHDand HEDT+2WHD,it can be seen that two upper impellers play a more important role on the change of RPD than the bottom impeller.

3.2.Volumetric mass transfer coefficient(kLa)

3.2.1.Influence of impeller combinations at various superficial gas velocities

kLa for different impeller combinations at various superficial gas velocities(uG)is shown in Fig.4.It can be seen that for a given specific power consumption,kLa changes only about 10%for five different impeller combinations when uGis small(e.g.uG=0.0039–0.0078 m·s-1).However,with uGincreasing from 0.016 to 0.039 m·s-1,the difference of kLa for five different impeller combinations increases gradually from 20%to as high as 80%.The impeller combination of PDT+2WHDhas the dominant mass transfer performance when uGranges from 0.0078 to 0.039 m·s-1in this work.It is also worth mentioning that the only up-pumping mode impeller combination HEDT+2WHUpresents much smaller mass transfer coefficient than other impeller combinations when uGis high(e.g.uG=0.031–0.039 m·s-1).

The regression results of kLa for different impeller combinations based on correlation(5)proposed by Cooper et al.[22]are shown in Table 2.

Fig.3.RPD for different impeller combinations at different agitation speeds.

Table 1 Regression results based on correlation(4)

Table 2 shows that the exponent β is bigger than α for all impeller combinations,indicating that it is more efficient to increase kLa by increasing the gas velocity than by increasing the power input,especially for the impeller combination of PDT+2WHD,the exponent β is twice as big as α.Compared with other impeller combinations,PDT as the bottom impeller has a better gas dispersion performance because of its lower power number and then higher agitation speed for a given power consumption.WHDas the upper impeller has a better effect to keep more bubbles in liquid because of its large blockage area and downpumping operation mode.Thus,the impeller combination PDT+2WHDcan lead to a higher gas holdup,lower RPD and better mass transfer performance than other impeller combinations,and this superiority becomes more evident with the increase of the gas velocity.Another impeller combination worth mentioning is HEDT+2WHU,which shows that the exponents α and β are relatively small,which can be ascribed to the up-pumping operation mode of the two top impellers.Min et al.[21]have simulated the flow field produced by HEDT+2WHUused in the present experiments with the result shown in Fig.5.The impellers discharge gas from the outermost edges of the impellers.The fluid at the level of the middle impeller is moving rapidly in a big Loop 2.This stream carries and accelerates upwards bubbles leaving the blade,thus reduces the residence time of bubbles in liquid,leading to a relatively high RPD,low gas holdup and bad mass transfer performance.

Fig.4.k L a for different impeller combinations and superficial gas velocities.

Table 2 Regression results based on correlation(5)

The detailed comparison and analysis of kLa for different impeller combinations will be discussed in the following sections.

3.2.2.Influence of bottom impeller

Fig.6 shows that when uGis in the range of 0.0039–0.0078 m·s-1,kLa for PDT+2WHDand HEDT+2WHDis almost the same,indicating that both PDT and HEDT can disperse gas wellatlow uG.Butathigher uGwhen the frequency of bubble collision and coalescence increases,kLa for PDT+2WHDis obviously higher than that for HEDT+2WHD.This is because the agitation speed ofPDT+2WHDis about1 s-1higher than that of HEDT+2WHDfor a given power consumption.Such as the specific power consumption Pgmequaling to 1.16 W·kg-1(corresponding to uG=0.039 m·s-1)in Fig.7,the agitation speed is 10 s-1for PDT+2WHDand 9 s-1for HEDT+2WHD.The stronger shear stress caused by the higher agitation speed can break the big bubbles into small ones more easily.Smaller bubbles can retain in liquid for longer time and increase the gas–liquid interfacial area,thus increase kLa eventually.

Fig.5.Predicted flow field of liquid stirred by HEDT+2WHU[21].

Fig.6.Comparison of k L a operated by PDT+2WHD and HEDT+2WHD.

3.2.3.Influence of mid and top impellers

Fig.8 shows the comparison of kLa obtained by impeller combinations PDT+2CBYN,PDT+2CBYWand PDT+2WHD,having different upper impellers with gradually increasing blade width.At low uGfrom 0.0039 to 0.0078 m·s-1,kLa is almost the same for all three impeller combinations;but at high super ficial gas velocity,kLa for impeller combination PDT+2WHDis 30%higher than that for PDT+2CBY,including PDT+2CBYNand PDT+2CBYW.The main reason is that the projection cross-sectional area of WHDis much bigger than that of CBY.The larger contact area of WHDon the liquid flow supplies more fluid circulation in the whole vessel,increasing the refresh frequency of the gas–liquid interfacial area and thus increasing kLa eventually.

Fig.7.Comparison of P gm-N for PDT+2WHD and HEDT+2WHD.

Fig.8.Comparison of k L a operated by PDT+2WHD and PDT+2CBY.

3.2.4.Influence of operating mode for same impeller

Fig.9 shows that when uGis in the range of0.0039–0.024 m·s-1,kLa for HEDT+2WHUand HEDT+2WHDis almost same for a given power consumption.As uGincreases,kLa for HEDT+2WHDis about 15%higher than that for HEDT+2WHU.With more gas dispersed,the bubble ascending velocity increases,the residence time of bubbles decreases,with a higher bubble escape velocity from the free surface.Fig.5[21]and Fig.10[23]show the flow field of liquid stirred by HEDT+2WHUand HEDT+2WHD,respectively.The liquid flow field produced by uppumping WHUwill accelerate the process of bubbles escaping while the down-pumping WHDcan decelerate the escaping process of bubbles.The opposite moving direction between gas and liquid flow produced by HEDT+2WHDrestrains bubble escape and increases the residence time of bubbles in liquid,then increases the gas holdup.Experimental results of Hao et al.[24]indicated that the gas holdup for HEDT+2WHDis obviously higher than that for HEDT+2WHU,and the exponent of uGin the gas holdup correlation for HEDT+2WHDis about twice of that for HEDT+2WHU.The increasing gas holdup will enlarge kLa;therefore,kLa for HEDT+2WHDis larger than that for HEDT+2WHU.

Fig.9.Comparison of k L a operated by HEDT+2WHU and HEDT+2WHD.

4.Conclusions

Five impeller combinations were used to study the effect of impeller combinations on gassed power and kLa at different superficial gas velocities in a baffled stirred reactor.

RPD is a vital characteristic in gas–liquid stirred tank design and operation.RPD decreases with increasing gas flow number(FlG)for all impeller combinations.Under all uGand rotation speeds,RPD of HEDT+2WHUis higher than that of all other combinations.That is mainly because the flow field of liquid produced by HEDT+2WHUhas the same direction of the gas flow,leading to a rapid release of bubbles from the liquid,thus to a reduction in the retained gas and so to a high average mixture density.

At low super ficial gas velocity,kLa for all impeller combinations under a given gassed power is almost the same.However,the impeller combination of PDT+2WHDshows an obviously superior mass transfer performance at high super ficial gas velocity.

Several independent factors affecting the mass transfer performance were analyzed,including bottom impeller,combination of two upper impellers,and the operating mode of WH impeller.As the bottom impeller playing an important role in gas dispersion,PDT has a higher agitation speed leading to a higher shear rate than HEDT for a given power consumption.The higher shear rate can break the big bubbles into small ones more easily,thus increase the gas–liquid interfacial area,and be beneficial to the mass transfer performance.Blade width is another important parameter affecting kLa.WH has a larger project area than CBYNand CBYW,offering a better gas–liquid flow circulation and increasing contact time and the interfacial refreshment rate for gas and liquid.Considering the superiorities of both the bottom impeller and the blade width of two upper impellers,PDT+2WHDis superior to other combinations,especially when the gas rate is high.In addition,the down pumping impeller produces downward axial flow to increase the bubble residence time,making PDT+2WHDattractive.

Finally,based on the experimental data,the regressed correlations of NPGwith the Froude number and gas flow number and that of kLa with specific power consumption and superficial gas velocity for five different impeller combinations were obtained and analyzed,providing helpful guidance in industrial stirred tank design.

[1]T.Moucha,V.Linek,E.Prokopova,Gas hold-up,mixing time and gas–liquid volumetric mass transfer coefficient of various multiple-impeller combinations:Rushton turbine,pitched blade and Techmix impeller and their combinations,Chem.Eng.Sci.58(9)(2003)1839–1846.

[2]M.Fujasová,V.Linek,T.Moucha,Mass transfer correlations for multiple-impeller gas–liquid contactors.Analysis of the effect of axial dispersion in gas and liquid phase on “l(fā)ocal”kLa values measured by the dynamic pressure method in individual sections of the vessel,Chem.Eng.Sci.62(6)(2007)1650–1669.

[3]L.Chen,Y.Y.Bao,Z.M.Gao,Void fraction distributions in cold-gassed and hotsparged three phase stirred tanks with multi-impeller,Chin.J.Chem.Eng.17(6)(2009)887–895.

[4]S.F.Yang,X.Y.Li,G.Deng,C.Yang,Z.S.Mao,Application of KHX impeller in a lowshear stirred bioreactor,Chin.J.Chem.Eng.22(10)(2014)1072–1077.

[5]Y.Y.Bao,J.Yang,B.J.Wang,Z.M.Gao,Influence of impeller diameter on local gas dispersion properties in a sparged multi-impeller stirred tank,Chin.J.Chem.Eng.23(4)(2015)415–422.

[6]J.M.T.Vasconcelos,S.C.P.Orvalho,A.M.A.F.Rodrigues,S.S.Alves,Effect of blade shape on the performance of six blade disk turbine impellers,Ind.Eng.Chem.Res.39(1)(2000)203–213.

[7]J.Zhao,Z.M.Gao,Y.Y.Bao,Effects of the blade shape on the trailing vortices in liquid flow generated by disc turbines,Chin.J.Chem.Eng.19(2)(2011)232–242.

[8]S.J.Arjunwadkar,K.Sarvanan,P.R.Kulkarni,A.B.Pandit,Gas–liquid mass transfer in dual impeller bioreactor,Biochem.Eng.J.1(2)(1998)99–106.

[9]H.Wu,V.Arcella,M.Malavasi,A study of gas–liquid mass transfer in reactors with two disk turbines,Chem.Eng.Sci.53(5)(1998)1089–1095.

[10]D.Garcia-Cortes,C.Xuereb,P.Taillandier,U.Jauregui-Haza,J.Bertrand,Effectof dual impeller-sparged geometry on the hydrodynamic and mass transfer in stirred vessels,Chem.Eng.Technol.27(9)(2004)988–999.

[11]M.Nocentini,D.Fajner,G.Pasquali,F.Magelli,Gas–liquid mass transfer and hold-up in vessels stirred with multiple Rushton turbines:water and water–glycerol solutions,Ind.Eng.Chem.Res.32(1)(1993)19–26.

[12]M.H.Xie,J.Y.Xia,Z.Zhou,G.Z.Zhou,J.Chu,Y.P.Zhuang,S.L.Zhang,Flow pattern,mixing,gas hold-up and mass transfer coefficient of triple-impeller configurations in stirred tank bioreactors,Ind.Eng.Chem.Res.53(14)(2014)5941–5953.

[13]M.S.Puthli,V.K.Rathod,A.B.Pandit,Gas–liquid mass transfer studies with triple impeller system on a laboratory scale bioreactor,Biochem.Eng.J.23(2005)25–30.

[14]X.Y.Li,G.Z.Yu,C.Yang,Z.S.Mao,Experimental study on surface aerators stirred by triple impellers,Ind.Eng.Chem.Res.48(18)(2009)8752–8756.

[15]V.Linek,T.Moucha,J.Sinkule,Gas–liquid mass transfer in vessels stirred with multiple impellers— I.Gas–liquid mass transfer characteristics in individual stages,Chem.Eng.Sci.51(12)(1996)3203–3212.

[16]N.Suhaili,J.S.Tan,M.Mohamed,M.Halim,A.B.Ariff,Effects of dual impeller system of Rushton turbine,concave disk turbine and their combinations on the performance of kojic acid fermentation by Aspergillus flavus Link 44-1,Asia Pac.J.Chem.Eng.10(2015)65–74.

[17]J.Markopoulos,C.Christo fi,I.Katsinaris,Mass transfer coefficients in mechanically agitated gas–liquid contactors,Chem.Eng.Technol.30(7)(2007)829–834.

[18]Y.Imai,H.Takei,M.Matsumura,A simple Na2SO3feeding method for kLa measurement in large-scale fermentors,Biotechnol.Bioeng.29(8)(1987)982–993.

[19]A.W.Nienow,M.Konno,W.Bujalski,Studies on three-phase mixing:a review and recent results,Chem.Eng.Res.Des.64(1)(1986)35–42.

[20]J.G.Long,Y.Y.Bao,Z.M.Gao,Gas–liquid dispersion in a stirred tank with different impeller combinations,J.Beijing Univ.Chem.Technol.32(5)(2005)1–5(in Chinese).

[21]J.Min,Y.Y.Bao,L.Chen,Z.M.Gao,J.M.Smith,Numerical simulation of gas dispersion in an aerated stirred reactor with multiple impellers,Ind.Eng.Chem.Res.47(18)(2008)7112–7117.

[22]C.M.Cooper,G.A.Fernstrom,S.A.Miller,Performance of agitated gas–liquid contactors,Ind.Eng.Chem.36(6)(1944)504–509.

[23]J.Zhao,X.N.Cheng,Z.M.Gao,Experimental study and numerical simulation of fluid flow in a liquid multiple impeller stirred tank,J.Beijing Univ.Chem.Technol.38(3)(2011)22–27(in Chinese).

[24]Z.G.Hao,Y.Y.Bao,Z.M.Gao,Gas–liquid dispersion in a multi-impeller stirred tank,J.Chem.Eng.Chin.Univ.18(5)(2004)547–552(in Chinese).