Shifang Yang,Xiangyang Li*,Gang Deng,Chao Yang*,Zaisha Mao

Fluid Dynamics and Transport Phenomena

Application of KHX Impeller in a Low-shear Stirred Bioreactor☆

Shifang Yang,Xiangyang Li*,Gang Deng,Chao Yang*,Zaisha Mao

Key Laboratory of Green Process and Engineering,Institute of Process Engineering,Chinese Academy of Sciences,Beijing 100190,China

A R T I C L EI N F O

Article history:

Bioleaching reactor

Draft tube

Mass transfer coeff i cient

Ore suspension

Shear stress

In our previous work,a low-shear stirred bioreactor was explored.With a pitched blade turbine impeller downf l ow(PBTD)used,the shear stress generated is high compared with that in some low shear axial f l ow impellers.KHX impeller is an eff i cient axial f l ow impeller,which provides large onf l ow diffusivity and low shear force.In this work,the KHX impeller wasappliedina lower-shearbioreactor and theperformance of this reactor was evaluatedand comparedwith that of thePBTDimpeller.Theexperimental results show thatthe KHX impeller can disperse gas at lower power consumption and gives greater gas-liquid volumetric mass transfer coeff icients than PBTD at the same power consumption.An empirical correlation for evaluating the mass transfer coeff i cient of the KHX impeller in the bioreactor is presented to provide reference for its industrial application. ?2014TheChemicalIndustry andEngineeringSocietyofChina,andChemicalIndustryPress.Allrightsreserved.

1.Introduction

In our earlier work[1],a low shear stirred tank bioreactor was designedbasedonthespecif i cfeaturescharacterizingbioleachingreactors andconventionalstirredtankreactors.Experimentalstudiesonsuspension of solid particles in gas-liquid-solid systems were conducted to simulate a real bioleaching system to examine the performance of this novel reactor.The results show that the f l ow f i eld in the reactor will be more uniform with a draft tube installed,reducing the maximal shear rate and the deviation from the average shear rate,which is benef i cial to a bioreactor.However,with the pitched blade turbine downf l ow(PBTD)impeller used,the average shear stress is high compared to that in some low shear axial impellers.

Based on the mechanism of solid suspension and gas dispersion,the solid suspension relies mainly on the turbulence generated by the impeller toward the bottom.When the downward discharge of theimpeller goes to the vessel bottom directly and turns toward the side walls, the liquid and solid particles are directed upwards to make particles suspend[2-5].For gas dispersion,radial impellers can offer a high rate of shear to liquid f l ow,which helps in dispersing the gas sparged into the reactor to smaller gas bubbles,increasing interfacial area between liquidandgasphases[6-8].Inagas-liquid-solidsystem,bothsolidsuspension and gas dispersion are important.In order to compromise the demand for low shear and large discharge,the ratios of power numbers andf l ownumbers,Np/Nqd,ofsomedifferenttypesofimpellersarecompared in Table 1.A high ratio means low discharge capacity and a low ratio represents low shear performance.

The KHX impeller(devised by Zhejiang Great Wall Reducer Co., Wenzhou,China)is an eff i cient axial f l ow impeller with power number of1andf l ownumberof0.75,whichcanprovidelargeonf l owdiffusivity and low shear force.In this work,a KHX impeller is installed in the previously tested lower-shear bioreactor[1]and the performance such as critical impeller speed,power consumption and gas-liquid volumetric mass transfer coeff i cient kLa of this reactor conf i guration is evaluatedandcomparedwiththatofusingthePBTDimpellerintheearlier work.

2.Experimental

2.1.Experimental setup

TheexperimentalsetupusedinthisstudyisshowninFig.1.Thebioreactor conf i guration is the same as that in[1]and has been depictedthere.The diameter(DT)of the bioreactor is 0.3 m and the height of the tank(HT)is 0.45 m.The water height(Hw)is equal to 0.42 m.Based on the rule of thumb that the cross-sectional area of channels is equal or nearly equal along the path of fl ow,the diameter of the draft tube is fi xedtobe0.2m.Theheightofthedrafttube(Hd)is0.31m,andthedistance(H1)betweenthetopofthedrafttubeandthefreesurfaceisabout 0.05 m before air sparging.A six-bladed 45°PBTD and a four-bladed KHX impeller downward pumping(as shown in Fig.2),both with diameter(DI)of 0.4 DT,are used in this study.Tap water,air and quartz are used as the liquid,gas and solid phases and the properties of solids and liquid are shown in Table 2.

Table 1Ratio of power numbers and f l ow numbers of different impellers[9,10]

Fig.2.Four-bladed KHX impeller.

Table 2Variations of solids and liquid

Fig.1.Experimentalsetup.1—bottleofNa2SO3solution;2—f l owmeter;3—stirredtank;4—drafttube;5—impeller;6—ringsparger;7—ovalbottom;8—oxygenprobe;9—dissolved oxygen meter;10—rotary torque transducer;11—motor;12—amplif i er;13—computer.

2.2.Methods of performance evaluation

2.2.1.Power consumption

The power consumed by the impeller was measured by a shafttorque method[11]using a dynamic torque sensor(model SN-1000, China Academy of Aerospace Aerodynamics,Beijing)installed between the impeller shaft and the gearbox.

2.2.2.Gas-liquid mass transfer

The gas-liquid volumetric mass transfer coeff i cient kLa was measured by the steady-state sulf i te feeding method[12].It is based on the measurement of dissolved oxygen concentrations under steadystate conditionswith theequilibrium between sulf i te additionand oxygen dissolution[13].The kLa value is calculated by

whereQsisthefeedrateofsul fi tesolution,Csistheconcentrationofsulfi te in the feed,and CAiis the equilibrium oxygen concentration in water,calculated according to Henry constant H and oxygen partial pressure.For the air-water system,CAican be calculated by

where CAis the oxygen concentration measured in water at steady state with the sulf i te solution fed continuously,and the Henry constant H for oxygen in water can be found in a physicochemical handbook.

In order to eliminate the inf l uence of temperature,the results are converted into the mass transfer coeff i cients at 20°C as follows:

2.2.3.Critical speed

Both NCDand Njsgwere measured by visual observation.Njsgis defi ned as the speed at which the solid particles deposit on the tank bottom is just eliminated[14]instead of the 1-2 s criterion proposed by Zweitering[15]and NCDis determined as the impeller speed above which all bubbles rise spirally around the shaft and small bubbles are uniformly distributed in the bulk fl ow[16].Several readings were taken to get the average for NCDand Njsg,and the error is about 1-2 r·min?1.

3.Results and Discussion

3.1.Critical impeller speed and power consumption

3.1.1.Critical impeller speed

For a three-phase stirred vessel bioreactor,simultaneous gas dispersion and solid suspension are very important.This requires that the impeller speed is greater than NCDand Njsg.Power consumption per volumetric liquid is an important index to the performance of a bioreactor.Since the differences between power numbers for different impellers may be more than 10 times[17],the power consumption of an agitator corresponding to a lower critical impeller speed can be possibly greater than that with higher critical impeller speed.In this study,the values of NCDand Njsgand the power consumption were measured with agitation by KHX and PBTD impellers (α=20%,Vs=7.35×10?4-2.21×10?3m·s?1)with or without a draft tube.

TheexperimentalresultsofNCDandNjsgareshowninTables3and4. Njsgof KHX is greater than that of PBTD while NCDis lower,because the fl ow patterns induced by PBTD and KHX are different.As discussed before,axial fl ow is more bene fi cial to suspending solids[17]and radial fl ow is more suitable for gas dispersion[6].Because PBTD and KHX are both axial impellers,the difference of Njsgis small.Since the curved blade of KHX facilitates the fl uid fl ow along the radial direction,the intensity of axial fl ow generated by KHX is weaker than that by PBTD,so that the values of Njsgwith KHX are greater than those with PBTD.For gas dispersion,the radial fl ow generated by KHX disperses gas sparging into the vessel to reach the wall easier,leading to lower values of NCDthan that by PBTD.

Considering that the power number of KHX is smaller than that of PBTD and the values of Njsgare higher than that of NCD,the power consumption for just suspending solids is calculated and the results are listed in Table 5.Even when the values of Njsgwith KHX are higher than those with PBTD,the power consumption of KHX is lower owing to a much smaller power number.By the criterion based on critical power consumption to characterizethe mixingin multiphase processes [18],KHX has more excellent performance for solid suspension and gas dispersion.

Table 3Results of NCDand Njsgwithout draft tube stirred by KHX and PBTD(α=20%)

Table 4Results of NCDand Njsgwith draft tube stirred by KHX and PBTD(α=20%)

Table 5Powerconsumption(W·m?3)corresponding to Njsgwith andwithouta draft tube stirred by KHX and PBTD(α=20%)

These results show that the critical impeller speed and power consumption are decreased obviously with a draft tube installed. This is because a draft tube can improve the f l ow pattern in a stirred reactor effectively.With a draft tube,the radial discharge from the impeller is restricted and the axial downward velocity is promoted, which favors to disperse solids and gas.Thus,an eff i cient top-tobottom circulation pattern is formed and the f l ow becomes more uniform[1].

3.1.2.Power consumption

As shown in Fig.3,power consumption decreases with increasing volumetric gas f l ow rates at a f i xed solid mass fraction.This is due to the formation of gas cavities behind the blades,which blocks transfer of energy from the agitator toward the liquid and the effect becomes stronger with increasing gas f l ow rates[19,20].The difference of power consumption between PBTDand KHX is shown in Fig.4.Because of the lower power number of KHX,the power consumption with KHX is much lower than that with PBTD under the same conditions and the effect of draft tube on power consumption for different impellers is the same,which would increase the power consumption in the agitator at a moderate rate.

Fig.3.Effect of gas f l ow rate on power consumption stirred by KHX(α=25%).■Vs= 7.35×10?4m·s?1;●Vs=1.47×10?3m·s?1;▲Vs=2.21×10?3m·s?1.

3.2.Gas-liquid mass transfer

3.2.1.Gas-liquid mass transfer coeff i cient

Fig.5 shows the experimental results of the gas-liquid volumetric mass transfer coeff i cient kLa in this low shear bioreactor stirred by theKHX impeller and the PBTD impeller.Because the impeller speed is higher than the critical impeller speed,the gas sparged through the agitator could be dispersed immediately and the gas holdup increases with gas f l ow rates,resulting in higher gas-liquid volumetric mass transfer coeff i cients.Stirred by the KHX impeller,greater values of kLa can be obtained at the same power consumption,because the KHX impeller has much wider impeller blades and breaks bubbles suff i ciently,increasing the gas-liquid interfacial area.On the other hand,the curved blades of the KHX impeller generate a radial f l ow pattern,which is conducive to gas dispersion.As shown in Fig.5,the values of kLa are higher with a draft tube because the local density of power consumption in the impeller zone is greater than that without a draft tube,leading to better gas dispersion.In conclusion,the KHX impeller has more excellent gas dispersion performance in such a stirred bioreactor.

3.2.2.Correlation of kLa

Many researchers have established empirical correlations in predicting the gas-liquid mass transfer coeff i cients in stirred tank reactors with different conf i gurations.The following type of correlation is often found in literature[21-25]:

Fig.4.Power consumption per volumetric liquid vs.impeller speed stirred by KHX and PBTD(α=20%,Vs=2.21×10?3m·s?1).■PBTD without draft tube;●KHX without draft tube;▲PBTD with draft tube;▼KHX with draft tube.

where P/VLis the power consumption per volumetric liquid (W·m?3),Vsis the superf i cial velocity of gas(m·s?1),and constants A,x and y depend on the conf i guration and geometry of the reactor.

As the values of exponent are different for the reactors with different geometries and impellers,a new correlation is established based on a regression analysis of the experimental data,which is suitable for the gas f l ow rate Vs=7.35×10?4-2.21×10?3m·s?1and solid mass fraction α=15%-25%.The correlation for the gasliquid mass transfer coeff i cients of the KHX and PBTD impellers in this low shear stirred bioreactor is as follows and the exponent values are shown in Table 6.

From the correlations it is known that the mass fraction of solids infl uences the gas-liquid mass transfer rate negatively[23].It seems that the solids added increase the viscosity of slurry,increasing the effect of bubble coalescence and lowering the specif i c area(a),further reducing kLa.It is noted from the large exponents for the power consumption per volumetric liquid that it has strong impact on the mass transfer coeff i cient.

Fig.5.Gas-liquidmasstransfervs.powerconsumptionwithdrafttubestirredbyKHXand PBTD(α=20%).■PBTDwithoutdrafttube;●KHXwithoutdrafttube;▲PBTDwithdraft tube;▼KHX with draft tube.

In order to demonstrate the signif i cance and accuracy of the correlation on mass transfer,the coeff i cients are recalculated throughregressive computation when exponents x,y and z are f i xed at their respective averages.The regression results in

Table 6Values of the exponents in Eq.(5)

and the mean relative deviations are shown in Table 7.It is more intuitive that KHX is more effective than PBTD as judged by kLa and a higher value of kLa can be obtained with a draft tube.

Table 7Regression for the coeff i cients in Eq.(6)with f i xed exponents

3.2.3.Correlation of power consumption

Power consumption is an important parameter to evaluate the mixingstatusina stirred tank.Many researchers have obtained correlations of power consumption for different impellers in single-and twophase systems[26,27].Dohi et al.[28]proposed a correlation for power consumption in a gas-liquid-solid three-phase reactor agitated by large-scale impellers:

where FrNis the Froude number and Frgis the Froude number for gas sparged.

BecauseoftheratiosofDIandDTforKHXandPBTDusedinthiswork are the same,the inf l uence of impeller diameter is not considered here and a newcorrelation of power consumption inthe presentgas-liquidsolid systems for KHX and PBTD is established:

whichissuitableforthegasf l owrateVs=7.35×10?4-2.21×10?3m·s?1and solid mass fraction α=15%-25%as shown in Table 8.

Table 8Regression coeff i cients in power consumption correlations

The predicted values by Eq.(8)and the experimental results are compared in Fig.6.The predictions f i t the experimental data well. Comparison of the exponent values in the correlations shows that the power consumption decreases with gas sparging and increases with stirring speed. 4.Conclusions

Fig.6.Comparison of P/VLin gas-liquid-solid three-phase systems for KHX and PBTD between experimental results and correlations.■PBTD without draft tube;●KHX without draft tube;▲PBTD with draft tube;▼KHX with draft tube.

In a low shear stirred bioreactor,the performance of KHX impeller was studied.The following conclusions are obtained.

(1)The KHX impeller could disperse gas at lower power consumption than PBTD in the same bioreactor.

(2)Greater values of kLa can be obtained with a draft tube at the same power consumption.

(3)Powerconsumptionintheagitatordecreaseswithincreasing gas fl ow rates and it is larger with a draft tube.

(4)The KHX impeller is more effective than PBTD as judged by kLa because of its excellent gas dispersion performance.

(5)The established empirical correlation can be used for evaluating the mass transfer coef fi cient and the power consumption of the KHX impeller in such a low shear stirred bioreactor,providing a reference for its industrial scale-up.

[1]X.Y.Li,C.Yang,G.J.Zhang,Z.-S.Mao,H.B.Zhou,Experimental studies on suspension of solid particles in a low shear stirred vessel,Chem.Eng.Technol.34(9)(2011) 1581-1586.

[2]M.Kraume,Mixing time in stirred suspensions,Chem.Eng.Technol.15(5)(1992) 313-318.

[3]W.Bujalski,K.Takenaka,S.Paolini,Suspensions and liquid homogenisation in high solids concentration stirred chemical reactor,Trans.IChemE 77(3)(1999)241-247.

[4]J.Wu,Y.Zhu,P.C.Bandopadhayay,Solids suspension with axial- fl ow impellers, AIChE J.46(3)(2000)647-650.

[5]I.Ayranci,M.B.Machado,A.M.Madej,J.J.Derksen,D.S.Nobes,S.M.Kresta,Effect of geometry on the mechanisms for off-bottom solids suspension in a stirred tank, Chem.Eng.Sci.79(2012)163-176.

[6]P.R.Gogate,A.A.C.M.Beenackers,A.B.Pandit,Multiple-impellers systems with a special emphasis on bioreactors:a critical review,Biochem.Eng.6(2)(2000)109-144.

[7]A.R.Khopkar,J.Aubin,C.Xuereb,Gas-liquid fl ow generated by a pitched-blade turbine:particle image velocimetry measurements and computational fl uid dynamics simulations,Ind.Eng.Chem.Res.42(21)(2003)5318-5332.

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

[9]R.J.Weetman,J.Y.Oldshue,Comparison of mass transfer characteristics of radial and axial fl ow impellers,Proc.6th European Conf.on Mixing,Pavia,Italy,1988, pp.24-26.

[10]J.H.Rushton,E.W.Costich,H.J.Everett,Power characteristics of mixing impellers, Chem.Eng.Prog.46(9)(1950)467-476.

[11]T.Wang,G.Z.Yu,Y.M.Yong,C.Yang,Z.S.Mao,Hydrodynamic characteristics ofdualimpeller conf i gurations in a multiple-phase stirred tank,Ind.Eng.Chem.Res.49(3) (2010)1001-1009.

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

[13]H.Wu,An issue on applications of a disk turbine for gas-liquid mass transfer,Chem. Eng.Sci.50(17)(1995)2801-2811.

[14]C.M.Chapman,A.W.Nienow,M.Cooke,Particle-gas-liquid mixing in stirred vessels:part III.Three phase mixing,Chem.Eng.Res.Des.61(3)(1983)167-181.

[15]T.N.Zweitering,Suspending solid particle in liquid by agitators,Chem.Eng.Sci.8 (3-4)(1958)244-253.

[16]Q.B.Cai,G.C.Dai,Flooding characteristics of hydrofoil impeller in a two-and threephase stirred tank,Chin.J.Chem.Eng.18(3)(2010)355-361.

[17]J.Wu,Y.Zhu,P.C.Bandopadhayay,L.Pullum,I.C.Shepherd,Solids suspension with axial-f l ow impellers,AIChE J.6(2)(2000)109-144.

[18]G.Deng,X.Y.Li,C.Yang,Z.-S.Mao,Critical power criterion for characterizing mixing in multiphase stirred vessels,Comput.Appl.Chem.29(4)(2012)383-386(in Chinese).

[19]N.Harnby,M.F.Edwards,A.W.Nienow,Mixing in the process industries,Powder Technol.47(1)(1986)101-102.

[20]R.P.Fishwick,J.M.Winterbottom,E.H.Stitt,Effect of gassing rate on solid-liquid mass transfer coeff i cients and particle slip velocities in stirred tank reactors,Chem. Eng.Sci.58(3-6)(2003)1087-1093.

[21]J.H.Rushton,J.J.Bimbinet,Holdup and f l ooding inair liquidmixing,Can.J.Chem.Eng. 46(1)(1968)16-21.

[22]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.

[23]B.?zbek,S.Gayik,The studies on the oxygen mass transfer coeff i cient in a bioreactor,Proc.Biochem.36(8-9)(2001)729-741.

[24]K.Van't Riet,Review of measuring methods and results in nonviscous gas-liquid mass transfer in stirred vessels,Ind.Eng.Chem.Process.Des.Dev.18(3)(1979) 357-360.

[25]V.Linek,P.Benes,J.Sinkule,Critical assessment of the steady-state Na2SO3feeding method for kLa measurement in fermenters,Biotechnol.Bioeng.35(8)(1990) 766-770.

[26]N.Dohi,Y.Matsuda,N.Itano,K.Shimizu,K.Minekawa,Y.Kawase,Mixing characteristics in slurry stirred tank reactors with multiple impellers,Chem.Eng.Commun. 171(1)(1999)211-229.

[27]Y.Kato,Y.Tada,Y.Takeda,Y.Hirai,Y.Nagatsu,Correlation of power consumption for propeller and pfaudler type impellers,J.Chem.Eng.Jpn 42(1)(2009)6-9.

[28]N.Dohi,T.Takahashi,K.Minekawa,Y.Kawase,Power consumption and solid suspension performance of large-scale impellers in gas-liquid-solid three-phase stirred tank reactors,Chem.Eng.Sci.97(2-3)(2004)103-114.

18 January 2013

☆Supported by the National Basic Research Program of China(2010CB630904), the National Natural Science Foundation of China(21276004,20990224),the National Natural Science Fund for Distinguished Young Scholars(21025627)and the National High Technology Research and Development Program of China (2012AA061503).

*Corresponding authors.

E-mail addresses:xyli@home.ipe.ac.cn(X.Li),chaoyang@home.ipe.ac.cn(C.Yang).

http://dx.doi.org/10.1016/j.cjche.2014.09.001

1004-9541/?2014 The Chemical Industry and Engineering Society of China,and Chemical Industry Press.All rights reserved.

Received in revised form 20 June 2013

Accepted 5 July 2013

Available online 6 September 2014