martinov_vlaev

Upload: kamlesh-sharma

Post on 30-Oct-2015

6 views

Category:

Documents


0 download

TRANSCRIPT

  • Increasing Gas-Liquid Mass Transfer in Stirred Power Law Fluids

    by Using a New Energy-Saving Impeller

    M. Martinov and S. D. Vlaev*Institute of Chemical Engineering, Bulgarian Academy of Sciences,Acad. G. Bonchev Str. Bl.103, 1113 Sofia, Bulgaria

    The gas-liquid mass transfer performance of a stirred vessel equipped with a novelimpeller, has been studied. The device is presented as means to increase gas-liquid masstransfer rate in stirred non-Newtonian media of fermentation vessels. The impeller is ofthe fluid foil type that is suitable for power law fluids. Mass transfer rates for power lawliquids are determined by measuring KLa in model pseudoplastic CMC and xanthan gumsolutions at moderate shear. The effects of geometry, gas flow rate, rheology, and impel-ler speed are illustrated. The results are compared with similar data reported for thedisc-style flat blade Rushton (RT) turbine, used conventionally in industrial vessels. Themass transfer coefficient is correlated with gas linear velocity and specific mixing power.An equation for KLa is proposed. In comparison with the conventional case, significantenergy saving is demonstrated.

    Key words:Mass transfer; stirred vessel; fluid foil impeller; pseudoplastic media

    Introduction

    Chemical and biochemical production technol-ogy is largely based on two-phase gas-liquid sys-tems with increased viscosity and complex-rheol-ogy. With regard to their diversity, such systems re-quire good mixing for homogenization, phase dis-persion and enhancement of mass transfer acrossthe interfaces. Because of the high viscosity andcomplex flow conditions, mixing in such systems iscarried out often by impeller-induced agitation.However, it is highly energy consuming. In orderto reduce power, while maintaining high perfor-mance, fluid foil impellers are employed.1 In addi-tion to the complex design2-5 of the blades, theseimpellers have also high solidity ratios, e.g. 0.9 to1.2. Due to their specific shape, they generally savepower.

    Our recent practice has led us to a new bladeshape (termed Narcissus, abbreviated NS) thatsaves considerable power.6, 7 This specific shape hasbeen found8 suitable for processing of non-Newto-nian fluids, as it ensures low degree of drag-reduc-tion generally associated with such systems.9 Onthe other hand, the gas hold-up in pseudoplastic dis-persions generated by this impeller, is relativelyhigh.10 Consequently, enhancement of mass transferhas been expected.

    In this paper, the mass transfer of the stirredsystem involving the new device is considered.

    Experimental

    Equipment

    The equipment employed (Fig.1) consisted ofstirred vessel 1 equipped with impeller 2, gas sup-ply lines 3, and measurement units 4 and 5. A baf-fled stirred vessel of standard geometry, i.e. D/Dv =0,33, H/Dv = 1 and h/D = 1, with diameter Dv =0,4 m, was used. Its liquid volume was 50 liters.

    Two types of impellers were used (Fig. 2): thespecific NS turbine and the conventional Rushton(RT) flat-blade disc style turbine. The NS-turbine isshown schematically in Fig. 2b. It contains7 of asupporting disc with attached even number of con-

    M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002) 1

    *To whom correspondence should be addressed. E-mail: [email protected]

    Original scientific paperReceived: November 7, 2001

    Accepted: February 5, 200

    F i g . 1 Experimental set-up

  • cave, equal-surface blades. Each pair of neighbourblades is positioned symmetrically over and underthe supporting disc plane. The blades are positionedin a way as to conclude acute angle with the sup-porting disc plane. The angle and blade shape canbe selected in a way as to ensure changes of, both,power consumption and gas dispersion capacity.

    In this paper, the mass transfer properties of awide-blade version with solidity ratio 0.9 suitablefor transitional mixing, are presented. Two versionsof NS with different blade angle , namely, 30 and60, were tested. The RT impeller has been used ascomparative basis.

    The gas supply lines contained gas cylinder fornitrogen, air compressor, flow rate meter, controlvalves and gas sparger. The gas sparger was a tubewith nine holes, each one of i.d. 0.8 mm. The gaswas introduced through the holes into the liquid bedunder the impeller at a distance equal to half of theimpeller off-bottom clearance.

    Model fluids

    Runs with aqueous solutions of carboxy-methylcellulose (CMC) and xanthan gum (XG),were accomplished. CMC was used as model of aweak pseudoplastic liquid, i.e. n > 0.6, and XG wasused as a strong pseudoplastic one, i.e. n > 0.5.

    Table 1 contains the viscometric parameters ofthe model liquid employed. For the shear stress

    and the apparent viscosity a in the Table, powerlaw was assumed, as follows: = K n and a =K n1, where = k ns stands for shear rate. TheMetzner constant k required in this equation hasbeen determined elsewhere.11 It has been found tobe k = 8.1 close to the value reported for axial flowdouble-pitch marine propellers.12

    Measurement technique

    The mass transfer rates were determined by thedynamic measurement technique. The studies in-volved preliminary tests of the probe response ingas phase to unit step change to confirm linearity ofthe measurement probe characteristic. The probe re-sponse in these experiments has been comparedwith the computed model response correspondingto the one-layer diffusion model of Lee and Tsao.13

    This way the probe response time was determinedto be 3 s in water. To measure KLa, the probe re-sponse in the liquid phase of the gas-liquid disper-sion to a step change in gas concentration was fol-lowed. In this case, the conventional procedure ofexperimental analysis has been fulfilled: first, thedissolved oxygen has been removed by nitrogensparging, then, a step change of gas concentrationhas been introduced by a sharp increase of oxygenflow rate using a pair of alternating magneticvalves. The dynamic probe response in the liquidphase

    = (R t R0) / (R s R0) (1)

    was evaluated by the MM (liquid and gas perfectlymixed) model14. In this relationship, R indicatesprobe readings at steady state (Rs), at t = 0 (R0), andat time t (Rt). KLa was determined as the slope of asemi-logarithmic plot ln (1 ) vs. time. Details ofthe specific analysis of the probe response are givenelsewhere.15

    Referring to Nocentini model analysis,14 modelselection has low impact for interpretation of exper-imental data in small-scale low-turbulence gas-liq-uid contactors. According to this analysis, modelMM produces less than 10 % error, provided thatKLa G H/vG < 0.1. The validity of this inequalityfor this study is apparent from the experimentaldata below. Nevertheless, in advance to our mainstudies in power law liquids, we checked the perfor-mance of the above technique in the case of masstransfer and mixing with a Rushton (RT) turbine inwater, for which system data can be found in the lit-erature. Our data for the system have been com-pared with reference data,16 e.g. the solid line inFig. 3. Two gas superficial velocities and nine im-peller speeds have been involved in this experi-ment. The compatibility of the measurement tech-

    2 M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002)

    T a b l e 1 Viscometric Data and Power-Law Quantitiesfor the Polymer Solutions

    FluidSystem

    Fraction

    w/%

    *

    Pa

    a*

    mPa.s

    K

    Pa.snn

    CMC 2 1.94 16.0 0.086 0.65

    Xanthan gum 0.5 6.84 56.4 2.27 0.23

    * Valid for = 121 s-1 at ns = 15 rps, 20 C

    F i g . 2 Impeller schematics: (a) RT (b) NS

  • nique has been confirmed for the important range ofspecific power 0.5 to 1.5 kWm-3.

    During the experimental runs, the torque andthe impeller revolutions, were controlled. Powerwas measured currently by a telemetrical systemElectroinventR. Viscosity and the flow curves weremeasured by a Haake rotational viscosimeter(Rheotest 2). On the other hand, gas hold-up wasmeasured locally by a conductivity probe and to-tally by the liquid displacement method.

    Results and Discussion

    Initially, in order to determine the effect of dif-ferent physical parameters, such as impeller geome-try, gas flow rate and power law coefficient, thevariation of KLa with impeller speed at various con-ditions, has been studied. Based on the specificpower input, some of these results were comparedfurther with similar data measured for the Rushtonturbine. Finally, a correlation for KLa has been pro-posed.

    Mass transfer rates versus impeller speed

    The effect of blade edge angle

    The effect of blade angle is illustrated in Fig.4. Mass transfer has been studied for the two impel-lers in two extreme cases, i.e. in water and in apower law xanthan gum solution. In both cases, therelative gas flow rate was 0.0004 m3 s1. One cansee that in the power law cases, the small angle is tobe preferred. In water, the difference between thetwo cases is not large. This result coincides with thegeneral concept of keeping low angles for hydro-foils (see Tatterson17). Consequently, the furtherstudies were based on the 30-degree version.

    Effect of gas flow rate and rheology

    As seen in Fig. 5, the effect of gas velocity isminor. And yet, one could see the general trend ofhigher KLa at a higher gas flow rate, due to the par-

    M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002) 3

    F i g . 3 Comparison of KLa-values obtained in this study (theexperimental. points) with reference values16 (thesolid line) for the important range (Rushton turbine)

    F i g . 5 KLa -versus ns for the NS-impeller in weak (a) andstrong (b) pseudoplastic dispersions at 0.0004and 0.0008 m3 s1, i.e. at vG 0.33 cm s

    1 and vG 0.66 cm s1, respectively

    F i g . 4 KLa-versus ns for the NS-impeller in a 30 degreeand 60 degree version

  • allel rise of gas hold-up in the gas dispersion equip-ment.

    The results for the weak and the strongpseudoplastic solutions are rather different. As seenin Fig. 6, the effect of the flow index, n, which is 1for water, 0.6 for CMC and 0.2 for the XG disper-sion, is well pronounced. A systematic decrease ofKLa is observed.

    In general, introducing polymer into a liquidleads to turbulence suppression plus intact interfa-cial area that increases the mass transfer resistance.On the other hand, the reduction of power draw, dueto polymer presence, lowers the gas hold-up and de-creases the interfacial area. Consequently, a drop inthe specific mass transfer rate occurs. However, it isessential to compare the rate of decrease to the per-formance of other well-known impellers, which isillustrated further.

    Comparison with the disc-style turbine

    The KLa - values determined for the NS turbineare compared with similar data obtained in thisstudy for the RT impeller at equal specific power(Fig. 7). Both weak and strong pseudoplastic fluidshave been tested. In the figure, the gas flow ratecorresponds to Q = 0.0004 m3 s1 for XG and to0.0008 m3 s1 for CMC.

    The plots in Fig. 7 show that the mass transfercoefficients determined for the NS impeller are gen-erally higher. The result can be explained by the gashold-up behaviour of the impeller. The gas hold-updata corresponding to the cases of Fig. 7 are com-pared in Fig. 8. One could observe large differencesin gas hold-up in favour of the fluid-foil impeller,

    4 M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002)

    F i g . 6 KLa vs. ns obtained for different rheology liquids:(u) water, (x) w = 2 % CMC solution, (s) 0.5 %XG solution

    F i g . 7 Comparison of KLa-values obtained for NS- andRT-agitated vessels in weak (a) and strong (b)pseudoplastic dispersions

    F i g . 8 Gas hold-up in the pseudoplastic solutions

  • especially in xanthan gum solutions, where the spe-cific power input exceeds 0.5 kW m3. This hasbeen explained10 by the uniformity of gas distribu-tion generated by this impeller which helps to avoidcavity formation and increases the gas contentaround the impeller and in the bulk.

    Correlation

    The results were correlated with respect to thegas superficial velocity, vG, and the specific powerdata, P/V, via equation (2):

    KLa = A(P/V)B vGC (2)

    h = 0.4 was assumed by reference to numerousstudies.16 Constants A and B were determined fromthe logarithmic plots of the experimental data. Thecorrelations for the NS turbine for CMC and XGare plotted in Fig. 9. One can see that the two lines,namely, the line corresponding to CMC and the onedescribing xanthan gum data, almost coincide witheach other.

    The following equations are proposed:for weak pseudoplastic:

    KLa = 0.26 (P/V)0.84 vG0.4, (3)

    for strong pseudoplastic:

    KLa = 0.23 (P/V)0.82 vG 0.4 (4)

    They are valid for

    0.1 < P/V < 2 kW m3

    3.310-3 m s-1 < vG < 6.610-3 m s1

    0.01 < < 0.1 Pas

    The corresponding ranges of Re, Po and Fl are:8

    1000 < Re < 25104; 0.5 < Po < 1; 0.01 < Fl < 0.04.From the relationships obtained one can see

    that the exponent at P/V is greater than exponent at

    vG. Consequently, the significance of the power ef-fect on mass transfer rate is great, i.e. B0.8. Thisresult coincides with previous observations due toAndrew18 and Kawase and Moo-Young19. The con-temporary survey for mass transfer in non-Newto-nian flow12 puts forward the empirical correlationfor KLa due to Kawase and Moo-Young19 . This cor-relation includes a very wide range of power-lawparameters. In this correlation, KLa (P/V)(9+4n)/10(1+n)

    vG0.5. Using the n-values of the CMC and XG solu-

    tions from this work, one finds exponents 0.7 and0.8 at P/V for CMC and XG, respectively. Thesevalues are close to the exponents determined above.

    Because the power law effect on gas-liquidmass transfer is included implicitly in the term ofP/V, we believe that, apart from the importance ofthe input power for the mass transfer rate, the largeexponent at P/V shows that the rheology effectupon KLa is high. However, here it has not been de-termined separately. On the other hand, the depend-ence of KLa on a found, most often19 is weak, e.g.KLa a

    0.25.Equations (3) and (4) can be used for interpola-

    tion within the range of 20 % relative error. Sincethe maximum difference of KLa obtained by theseequations reaches 15 % comparable with the maxdeviation of experimental error of 20 %, the differ-ence of parameters A in these equations could beassumed insignificant, so that only one of theseequations is used for both fluids. However, usingseparate equation for CMC and XG could be justi-fied in some ranges of KLa, where the experimentalerror is lower.

    As estimated from equations (3) and (4), andFigs. 7a and 7b, the power per unit volume requiredby the NS impeller to reach a KLa value similar tothe Rushton turbine, is lower. For example, in orderto reach KLa 0.02 s1 in weak psedoplastic me-dium, RT requires 1 kW m3, while NS requires 0.4kW m3 (see Fig. 7a). On the other hand, to reachKLa 0.01 s1 in strong pseudoplastic fluid the val-ues are 1.5 kW m3 and 0.4 kW m3 for the RT andNS, respectively (see Fig. 7b). Consequently, thesame mass transfer rate is obtained by the fluid-foilimpeller with a significant energy saving. This factmay look strange on the background of the generalconclusion that equal power per unit volume andsuperficial gas velocity leads to the same KLaregardless of the impeller type20. However, thisis true for developed turbulent flow conditions. Innon-Newtonian flow and transient flow conditionscorresponding to higher viscosity, a low powernumber agitator is driven at a higher speed to givethe same specific power and thus gives higher KLabecause of the lower apparent viscosity achieved.Regarding recent computational fluid dynamic(CFD) analyses of the NS impeller21, there is one

    M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002) 5

    F i g . 9 Correlation of gas-liquid mass transfer in powerlaw liquids for the NS impeller

  • more reason for producing high gas hold-up andKLa in complex-rheology fluids by means of thisimpeller. In polymer presence, the NS turbine ex-hibits lower degree of drag reduction and flat powercharacteristics compared to the RT and power inNewtonian flow.

    Conclusions

    The study reveals the gas-liquid mass transferproperties in power law media of a hitherto un-known impeller configuration. It is confirmed, thatusing impellers with improved fluid-foil blades instirred vessels, mass transfer could be increased upto 100 % compared to conventional bioreactorsequipped with conventional turbines. The increaseis believed to be due to the greater uniformity ofgas hold-up and hindrance of the power drag reduc-ing property of the impeller-liquid system causedby the specific fluid-foil blade shape NS. Comparedto the conventional system, the NS turbine exhibitsalso significant energy saving.

    ACKNOWLEDGEMENTS

    The study has been supported partially by Eu-ropean Commission Grant No. IC15CT980502. Theanalysis of the 60-degree version of NS has beencontracted by the Bulgarian National ScienceFoundation Grant No. TN-710 / 97.

    S y m b o l s

    h impeller off-bottom clearance, mD impeller diameter, mH liquid height, mK consistency coefficient, Pa sn

    n flow behaviour indexns impeller speed, s

    1

    KLa volumetric mass transfer coefficient, s1

    P power, WP/V specific mixing power, W m3

    Dv vessel diameter, mQ volumetric flow rate, m3 s1

    v superficial gas velocity, m s1

    V liquid volume, m3

    w mass fraction, %

    G r e e k l e t t e r s

    shear rate, s1

    phase hold-up density, kg m3

    , a viscosity, apparent viscosity, Pa. s shear stress, Pa

    D i m e n s i o n l e s s n u m b e r s

    Fl flow number, Q/ns D3

    Re Reynolds number, ns D2 /

    Po power number, P/ns3 D5

    I n d i c e s

    G gas phase

    L liquid liquid phase

    R e f e r e n c e s

    1. Oldshue, J. Y., Wheetman, R. J., Proceedings of the 6th Eu-ropean Conference on Mixing, BHRA Fluid Engineering,Pavia/Italy, 1988, pp 4345.

    2. Oldshue, J. Y., Chem Eng. Prog. 50 (1989) 33.

    3. Fort, I., Medek, J., Ambros, F., Chem. Biochem. Eng. Q.13 (1999) 127.

    4. Mishra, V. P., Dyster, K. N., Jaworski, Z., Nienow, A.,McKemmie, J., Can. J. Chem. Eng. 76 (1998) 577.

    5. EKATO, Process Improvement with a novel gassing im-peller. EKATO Mitteilung No. 18, pp 18, 1998.

    6. Vlaev, S. D., Martinov, M., Impeller for mixing com-plex-rheology media in the process industries. In Proceed-ings of ACHEMA Session on Mechanical Process Engi-neering, DECHEMA, Frankfurt am Main, 1997, pp.103105.

    7. Kraitschev, St., Lossev, V., Vlaev, S. D., Valeva, M., Inst.Chem. Eng. Symp. Ser. No. 146 (1999) 245.

    8. Vlaev, S. D., Martinov, M., Inst. Chem. Eng. Symp. Ser.No. 146 (1999) 253.

    9. Harnby, N., Edwards, M. F., Nienow, A. W., Mixing in theProcess Industries, Butterworth, London, 1992.

    10. Vlaev S. D., Martinov M., Avoiding pathological gas dis-persion by using a hydrofoil double impeller. In Proceed-ings of ACHEMA Session on Bioprocess Engineering,DECHEMA, Frankfurt am Main, 2000, pp. 7376.

    11. Martinov, M., Vlaev, S. D., Bulg. Chem. Commun. 31(2000) 471.

    12. Chhabra, R. P., Richardson, J. F., Non-Newtonian Flow inthe Process Industries, Butterworth-Heinemann, Oxford,1999.

    13. Lee, Y. H., Tsao, G. T., Dissolved oxygen electrodes. InFiechter A. (Ed.), Adv. Biochem. Eng., Vol. 13, pp 3586,Springer Verlag, Berlin, 1979.

    14. Nocentini, M., Trans. Inst. Chem. Eng. 68A (1990) 287.

    15. Vlaev, S. D., Valeva, M., J. Biotechnol. 11 (1989) 83.

    16. Linek, V., Vacek, V., Benes, P., Chem. Eng. J. 34 (1987) 11.

    17. Tatterson, G. B., Fluid Mixing and Gas Dispersion in Agi-tated Tanks, McGraw-Hill, NewYork, 1991.

    18. Andrew, S. P. S., Trans. Inst. Chem. Eng. 60 (1982) 3.

    19. Kawase, Y. and Moo-Young, M., Chem. Eng. Res. Des. 66(1988) 284.

    20. Nienow A. W., Trans. Inst. Chem. Eng. 74A (1996) 417.

    21. Vlaev, S. D. , Staykov, P., Analysis of the Drag-ReducingEffect of Blade Shape on Impeller Performance in StirredPower Law Fluids. In Proceedings of the Balkan Seminaron Rheology, Bulgarian Academy of Sciences, Sofia,2001, pp. 138146.

    6 M. MARTINOV and S. D. VLAEV, Increasing Gas-Liquid Mass Transfer , Chem. Biochem. Eng. Q. 16 (1) 16 (2002)