Optimal Operation of an Integrated Bioreaction–Crystallization Process

中国科技论文在线 http://www.paper.edu.cn

Chemical Engineering Science 56 (2001) 6165–6170 www.elsevier.com/locate/ces

Optimal operation ofan integrated bioreaction–crystallization process for continuous production of calcium gluconate using external loop airlift columns Jie Baoa, Kenichi Koumatsua, Keiji Furumotob, Makoto Yoshimotoa, Kimitoshi Fukunagaa, Katsumi Nakaoa; ∗

aDepartment of Applied Chemistry and Chemical Engineering, Yamaguchi University, Tokiwadai, Ube, Yamaguchi 755-8611, Japan bOshima National College of Maritime Technology, Oshima, Yamaguchi 742-2106, Japan

Abstract A kinetic model was proposed to optimize the integrated bioreaction–crystallization process newly developed for production of calcium gluconate crystals using external loop airlift columns. The optimal operating conditions in the bioreactor were determined using an objective function deÿned to maximize the productivity as well as to minimize biocatalyst loss. The optimization of the crystallizer was carried out by matching the crystallization rate to the optimal production rate in the bioreactor because the bioreaction was found to be the rate controlling process. The calcium gluconate productivity under the optimal conditions of the integrated process was obtained by the simulation based on the process model. The productivity ofthe proposed process was found to be comparable to that ofthe current batch fermentationprocess. ? 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Process kinetic model; Optimization; Calcium gluconate; Bioreaction–crystallization process; Immobilized glucose oxidase; Biocatalyst deactivation

1. Introduction crystallization kinetics were studied in the authors’ pre- vious works (Nakao et al., 1991; Nakao, Kiefner, Furu- In this work, a new environment friendly bioprocess moto, & Harada, 1997; Nakao, Bao, Harada, Yasuda, & for the continuous production of calcium gluconate crys- Furumoto, 2000a; Nakao et al., 2000b; Nakao, Koumatsu, tals has been proposed using the immobilized glucose ox- Bao, Yoshimoto, & Fukunaga, 2000c; Bao, Furumoto, idase catalyzed oxidation ofglucose. The new integrated Fukunaga, & Nakao, 2000). The purpose ofthis study bioreaction–crystallization process employs two external is to develop an overall process model for the integrated loop airlift bubble columns in series, one being bioreac- bioreaction–crystallization process, and then to optimize tor and the other crystallizer. The bioreactor is used to the operating conditions through the objective function produce calcium gluconate from the oxidation of glucose deÿned so as to obtain the maximum productivity at the catalyzed by the immobilized glucose oxidase plus man- minimum loss ofbiocatalyst. Finally, the advantages of ganese dioxide gel beads. The output from the bioreactor the new process have been demonstrated by comparing it is then fed into the crystallizer to produce the calcium with the current batch fermentation process (Milsom & gluconate crystals. The enzymatic reaction kinetics, the Meers, 1985; Rao & Panda, 1993). 9ow and mixing behaviors ofthe immobilized glucose oxidase gel beads, the gas–liquid and liquid–solid oxygen transfer properties, the deactivation kinetics of the immo- 2. Process development and design equations bilized glucose oxidase plus manganese dioxide, and the 2.1. Reaction scheme and process 0owsheet ∗ Corresponding author. Tel.: +81-836-85-9271; fax: +81- 836-85-9201. The reaction scheme for production of calcium glu- E-mail address: [email protected] (K. Nakao). conate is composed ofthree separate reactions, the

0009-2509/01/$ - see front matter ? 2001 Elsevier Science Ltd. All rights reserved. PII: S 0009-2509(01)00272-X 转载 中国科技论文在线 http://www.paper.edu.cn 6166 J. Bao et al. / Chemical Engineering Science 56 (2001) 6165–6170

 immobilized glucose oxidase catalyzed oxidation ofglu- CE and CM as well as eKectiveness factors  and  , ac- cose (1), the hydrogen peroxide decomposition (2) cat- cording to Eqs. (10)–(13). alyzed by the immobilized manganese dioxide, and the (2) Deactivation kinetics of immobilized glucose oxidase neutralization ofgluconic acid with calcium hydroxide and manganese dioxide (Nakao et al., 2000b):

(3). The overall reaction is represented by reaction (4) −5 under the steady state ofoxygen and hydrogen peroxide −(dCE=dt)=kdCE =(9:79 × 10 [pH] concentrations: −4 −3:79 × 10 )CPS CE; (10) Glucose(G)+O + H O 2 2  −3 −(dCM =dt)=kdCM =(1:13 × 10 GO →Gluconic acid(GA)+H2O2(P); (1) −4 −1:57 × 10 [pH])CPS CM : (11)  MnO2 1 (3) Correlations of initial  and  k values (Nakao H2O2(P) → H2O + O2; (2) P 2 et al., 1991): 1 ln  = ln(9:890 × 10−3[pH] − 3:149 × 10−2) Gluconic acid(GA)+ 2 Ca(OH)2 1 −0:405 ln CG − 0:244 ln CE; (12) → 2 Calcium gluconate(CaG)+H2O2; (3)

 −3 −2 1 1  kP =63:42CM − 9:024 × 10 [pH]+7:104 × 10 : Glucose(G)+ O2 + Ca(OH)2 2 2 (13) → 1 Calcium gluconate(CaG)+H O: (4) 2 2 (4) Correlations of mass transfer properties with oper- The schematic ofthe proposed process is shown in ating parameters (Bao et al., 2000): Fig. 1. The highly concentrated calcium gluconate solu- k a =2:567 × 10−2 +1:021 ln U ; (14) tion produced in the bioreactor is fed into the crystallizer ln L G in which the product is crystallized for its recovery. The mother liquid ofthe crystallizer is recycled to the biore- ln kS = − 4:018+0:276 ln UG: (15) actor. Hence, the feed of solid glucose and calcium hy- (5) Crystallization kinetics of calcium gluconate from droxide to the bioreactor with simultaneous removal of reaction solution (Nakao et al., 2000c): calcium gluconate crystals from the crystallizer realizes a continuous production ofthe crystals with almost no dCCaG=dt =2CS (1=LS +1=rS )Kg[CCaG − 2:61 waste eIuents. ×105 exp(−2:64 × 103=T)]3:16; (16)

2.2. Design equations where the crystallization rate constant Kg =(6:35 × 10−12[pH] − 2:47 × 10−11) exp(−8:71 × 103=T). (1) Mass transfer and reaction kinetic equations (Nakao et al., 1991): 2.3. Objective function for overall process optimization dC =dt =(−dC =dt)=2 (5) CaG G The objective function was deÿned as the ratio of the calcium gluconate productivity to the catalyst cost dur- = kLa(COi − COS ) (6) ing the operation period equal to half-life of the immo- bilized glucose oxidase CE;t1=2(= ln 2=kd) as shown in ∗ Eq. (17). The cost consists ofthe immobilized glucose = kS aS (COS − C )(VS =VL) (7) OS oxidase and manganese dioxide deactivated per calcium ∗ gluconate produced. The unit ofthe objective functionis 1 VM C VS OS 2 3 = ∗ (8) mol =(m $). 2 COS + KM (1 + CPS =KI ) VL Objective = k C (V =V )=2; (9) P PS S L
    t1=2 dCCaG where the apparent Michaelis constant KM = kmO=(1 + = dt 0 dt kmG=CG), the maximum reaction rate VM = kCATCE=(1 +
 t     t   kmG=CG), the competitive inhibition constant KI which 1=2 dC 1=2 dC P − E dt CaG dt was found to be almost equal to KM . The three unknown GO dt dt variables C , C∗ , and C can be determined by rear- 0 0 OS OS PS         ranging the equations, Eqs. (7)–(9), taking into account t1=2 dC t1=2 dC + P − M dt CaG dt the progressive change ofthe catalyst concentrations MnO2 0 dt 0 dt 中国科技论文在线 http://www.paper.edu.cn J. Bao et al. / Chemical Engineering Science 56 (2001) 6165–6170 6167

Fig. 1. Schematic diagram ofintegrated bioreaction–crystallization process forproduction ofcalcium gluconate crystals.


   2
   t1=2 t1=2 dCCaG dCE 5 to 6 (Nakao et al., 1991). In the immobilized glucose = dt − dtPGO oxidase catalyzed reaction, the same optimal temperature 0 dt 0 dt of30 ◦C was also adopted. However, the optimal pH value     t1=2 dC had to be changed because the immobilized manganese + − M dtP ; (17) dt MnO2 dioxide reacted with the produced hydrogen peroxide in 0 the acidic solution. Fig. 2 shows the eKect ofpH value on where P and P are the molar prices ofglucose GO MnO2 the deactivation ofthe immobilized glucose oxidase and 6 oxidase (PGO is $9 × 10 =mol) and manganese dioxide manganese dioxide. The lower pH was favorable for the (P is $12.5=mol), respectively. Maximization ofthe MnO2 enzyme stability as shown in Fig. 2(a). On the other hand, objective function means the maximum production of cal- Fig. 2(b) shows that the immobilized manganese dioxide cium gluconate at the minimum loss ofbiocatalyst in the deactivates faster at the lower pH value. To perform the bioreactor. Because the bioreaction was found to be the prolonged practical operation, therefore, pH 7 was taken rate controlling step for the overall process, the optimiza- as the optimal one in spite ofa lower enzyme stability. tion ofthe crystallizer could be achieved by matching the crystallization rate to the optimal production rate in the bioreactor. Thus, it is easier and more 9exible to carry 3.2. Determination of optimal conditions in bioreactor out the calculation in a separate calculation tool than to with objective function use the process simulation system like Aspen or Hysys. Aspen or Hysys is very eNcient on the 9owsheet simu- The glucose concentration CG, the immobilized glu- lation and optimization, but shows no advantage for the cose oxidase concentration CE, the manganese dioxide optimization calculation in a single special operation unit concentration CM , the gel beads loading S , and the su- which is not the standard modular ofthe package. Be- perÿcial gas velocity UG were changed in their practical sides, the calculation using Aspen or Hysys needs to build range to maximize the objective function. many Fortran blocks to describe the particular reaction Fig. 3(a) shows that the objective function decreases and deactivation kinetics, mass transfer behaviors etc. with increasing CE. This suggests that the lowest possi- ble CE is most favorable for the maximum productivity with the minimum biocatalyst loss, because the lower CE 3. Results and discussion is favorable for the less accumulation of hydrogen per- oxide. Nevertheless, the optimal CE was determined as 3.1. Determination of optimal temperature and pH 1:0×10−3 mol=m3, considering a diNculty in an accurate value in bioreactor determination ofthe lower CE and a practical production rate ofcalcium gluconate. The kinetic results on the free glucose oxidase cat- With an increase in CG, the intrinsic oxidation rate in- alyzed reaction showed that the optimal reaction tem- creases but the eKectiveness factor decreases (Eqs. (5)– ◦ perature was 30 C and the optimal pH value was from (7) and (12)). On the other hand, increasing CG led to an 中国科技论文在线 http://www.paper.edu.cn 6168 J. Bao et al. / Chemical Engineering Science 56 (2001) 6165–6170

Fig. 2. EKects ofpH value on kd=CPS for immobilized enzyme (a) and kd=CPS for immobilized manganese dioxide (b) under various glucose and hydrogen peroxide concentrations. increases in CPS (Eqs. (7) and (8)) and hence the deac- tivation rate ofenzyme (Eq. (10)). Fig. 3(b) shows that the objective function initially increases with increasing 3 CG, passes a maximum value at 8 kg=m , and then de- creases steadily. Thus, the optimal CG was determined as 8 kg=m3. Increasing UG caused an increase in the production Fig. 3. Determination ofoptimal operating conditions in bioreactor. rate due to the increasing oxygen transfer rate and did lit- Dotted lines indicate optimal conditions. tle change in both deactivation rates ofbiocatalyst. The reason for the latter fact might be that the hydrogen per- In summary, the optimal biocatalyst compositions oxide mass transfer rate was enhanced by increasing UG (Eqs. (12) and (13)), leading to a decrease in C (Eqs. and operating conditions in the bioreactor were deter- PS −3 3 (6)–(9)) and hence both deactivation rates (Eqs. (10) mined as follows: CE =1:0 × 10 mol=m , CM =8wt%, 3 and (11)). Fig. 3(c) shows that the objective function in- CG =8kg=m , S =0:40 and UG =0:04 m=s. creases with increasing UG. At the highest UG of0 :04 m=s within the range concerned, the function is seen to reach 3.3. Determination of optimal conditions in crystallizer almost its maximum value. Hence, the optimal UG was determined as 0:04 m=s. The bioreaction rate in the ÿrst airlift column was Although the high CM contributes little to the produc- found to be the rate controlling step in the integrated tion rate directly, it leads to a decrease in CPS (Eq. (9)) bioreaction–crystallization process. Therefore, the opti- and hence both deactivation rates ofbiocatalyst (Eqs. mization ofthe crystallizer should be achieved by match- (10) and (11)). The higher S brings about the higher ing the crystallization rate ofcalcium gluconate to the op- production rate as well as the lower deactivation rates timal production rate determined above in the bioreactor. ofbiocatalyst (Eqs. (7)–(11)). Thus, the highest CM of Fig. 4(a) shows that the crystallization rate increases 8 wt% and the highest S of0.40 within the range exam- with increasing pH in the pH range of4–7 according to ined were taken as their optimal values, respectively. Eq. (16). The crystallization rate only at pH 7 is seen 中国科技论文在线 http://www.paper.edu.cn J. Bao et al. / Chemical Engineering Science 56 (2001) 6165–6170 6169

Fig. 5. Time courses ofreaction and deactivation rates under optimal conditions in prolonged continuous operation. t1=2 and t1=2 mean the half-life of immobilized glucose oxidase, CE , and that ofcalcium gluconate productivity, dCCaG=dt, respectively.

for both immobilized glucose oxidase and manganese dioxide were simulated by solving the design equations, Eqs. (5)–(16) simultaneously using the Runge–Kutta nu- merical method. The results shown in Fig. 5 indicate that the calcium gluconate productivity decreases much slower than the deactivation rates ofboth catalysts do. Fig. 4. Determination ofoptimal operating conditions in crystallizer. Dotted lines indicate the maximum productivity ofcalcium gluconate This is due to the fact that oxygen and hydrogen peroxide in bioreactor. transfer rates within the gel beads control their overall reaction rates. The calcium gluconate productivity can be evaluated to meet the optimal productivity in the bioreactor in the in terms ofthe glucose consumption rate. Thus, the aver- range ofthe crystal concentration CS concerned. Further- age glucose consumption rates during the operation pe- more, since pH 7 is also the optimal pH value in the riod equal to the half-life t1=2 ofimmobilized glucose ox- 3 bioreactor, no further pH adjustment is needed in the re- idase CE, was calculated as 2:9 kg=(m h). On the other cycling operation between two columns. hand, the average rate during the period equal to the  Fig. 4(b) shows the variation ofthe crystallization rate half-life t1=2 ofcalcium gluconate productivity dCCaG=dt, ◦ ◦ with the temperature from 5 C to 25 C at pH 7 based was calculated as 2:3 kg=(m3 h). Such performances can on Eq. (16). The results show that the rate at 10◦C is be compared with the reported glucose consumption rate the highest one irrespective ofthe CS value. Thus, the in the current batch fermentation process, which was ◦ temperature of10 C was taken as the optimal one in the 4:3 kg=(m3 h) only in the batch fermenter section (Mil- crystallizer. Fig. 4(b) also indicates that 47 kg=m3 is the som & Meers, 1985). The fermentation process requires minimum CS value to meet the optimal production rate much time for the upstream treatment such as prepara- in the bioreactor. tion ofnutrients, inoculum and sterilization as well as the In summary, the optimal operating conditions in the downstream treatment. Consequently, the overall produc- ◦ crystallizer were determined as pH 7, 10 C and CS of tivity in the proposed process is at least comparable to 47 kg=m3. that ofthe batch fermentationprocess. In addition, the merits ofthe less waste eIuents and simpler process also 3.4. Comparison between proposed process and batch support that the proposed process is superior to the batch fermentation process fermentation process.

The calcium gluconate productivity and glucose con- sumption rate decrease gradually in the prolonged contin- 4. Conclusion uous operation because ofthe deactivation ofbiocatalyst. The time courses ofthe calcium gluconate productivity, (1) A process model for the integrated bioreaction– the glucose consumption rate, and the deactivation rates crystallization process using external loop airlift columns 中国科技论文在线 http://www.paper.edu.cn 6170 J. Bao et al. / Chemical Engineering Science 56 (2001) 6165–6170 was proposed to optimize production ofcalcium glu- t1=2 half-life of immobilized glucose oxidase, h or conate crystals. The optimal operating conditions in the day bioreactor were determined based on an objective func-  t1=2 half-life of calcium gluconate productivity, h tion deÿned so as to maximize the calcium gluconate pro- or day ductivity as well as to minimize the biocatalyst loss. UG superÿcial gas velocity, m=s 3 (2) The optimization ofthe crystallizer could be carried VL liquid volume, m out by matching the crystallization rate to the optimal VM apparent maximum reaction rate for free glu- production rate in the bioreactor. cose oxidase, mol=m3 · s 3 (3) The average calcium gluconate productivities un- VS gel beads volume, m der the optimal conditions for the integrated process were obtained by the simulation based on the process model. The productivity ofthe proposed process was foundto Greek letters be comparable to that ofthe current batch fermentation  eKectiveness factor for immobilized glucose process. oxidase  eKectiveness factor for immobilized man- ganese dioxide  Notation  kP apparent ÿrst-order rate constant for immobi- lized manganese dioxide, 1=s aS speciÿc surface area of gel beads, 1=m 3 S fractional gel beads loading in bioreactor CCaG concentration ofcalcium gluconate, kg=m CE concentration ofimmobilized glucose oxi- dase, mol=m3 References 3 CG concentration ofglucose, kg=m CM concentration ofimmobilized manganese Bao, J., Furumoto, K., Fukunaga, K., & Nakao, K. (2000). Average dioxide, wt% and local oxygen transfer properties in bubble column with axial distribution ofimmobilized glucose oxidase gel beads. Chemical COS steady state concentration ofdissolved oxy- gen, kg=m3 Engineering Science, 55, 5405–5414. ∗ Milsom, P., & Meers, J. (1985). Gluconic and itaconic acids: In COS steady state concentration ofdissolved oxy- comprehensive biotechnology. Vol. 3. (pp. 681–700), Oxford: gen on the gel beads surface, kg=m3 Pergamon Press. CPS steady state concentration ofhydrogen perox- Nakao, K., Fukunaga, K., Yasuda, Y., Furumoto, K., Ohtani, S., & ide, kg=m3 Kimura, M. (1991). Measurement ofoxygen transferproperties C crystal concentration in crystallizer, kg=m3 using oxidation ofglucose with air catalyzed by glucose oxidase. S Kagaku Kogaku Ronbunshu, 17, 873–881. kCAT rate constant for free glucose oxidase, 1=s Nakao, K., Kiefner, A., Furumoto, K., & Harada, T. (1997). kd ÿrst-order rate constant ofimmobilized glu- Production ofgluconic acid with immobilized glucose oxidase in cose oxidase deactivation, 1=s airlift reactors. Chemical Engineering Science, 52, 4127–4133.  Nakao, K., Bao, J., Harada, T., Yasuda, Y., & Furumoto, K. (2000a). kd ÿrst-order rate constant ofimmobilized man- ganese dioxide deactivation, 1=s Measurement and prediction ofaxial distribution ofimmobilized glucose oxidase gel beads suspended in bubble column. Journal kLa volumetric gas–liquid oxygen transfer coeN- of Chemical Engineering of Japan, 33, 721–729. cient, 1=s Nakao, K., Koumatsu, K., Bao, J., Furumoto, K., Yoshimoto, M., kmG Michaelis constant with respect to glucose for & Fukunaga, K. (2000b). Deactivation kinetics ofimmobilized free glucose oxidase, mol=m3 glucose oxidase and manganese dioxide for production of calcium gluconate in an airlift reactor, Preprints of 33rd autumn meeting kmO Michaelis constant with respect to oxygen for 3 of SCEJ, Shizuoka, Japan, September (p. 305). free glucose oxidase, mol=m Nakao, K., Koumatsu, K., Bao, J., Yoshimoto, M., & Fukunaga, kS liquid-solid mass transfer coeNcient around K. (2000c). Kinetics ofcrystallization ofcalcium gluconate from gel beads, m=s reaction solution for production of calcium gluconate in an airlift reactor, Preprints of Himeji meeting of SCEJ, Himeji, Japan, KI competitive inhibition constant ofhydrogen peroxide to glucose oxidase, mol=m3 July (pp. 16–17). Rao, D. S., & Panda, T. (1993). Comparative analysis ofcalcium KM apparent Michaelis constant for free glucose gluconate and sodium gluconate techniques for the production of oxidase, mol=m3 gluconic acid by Aspergillus niger. Bioprocess Engineering, 8, t time, s or h 203–207.