节点文献
多层搅拌式生物反应器内溶液流变性质对流场特性影响的研究
Study of the Effect of the Rheology Properties on Flow Fields in Stirred Bioreactors with Multiple Impellers
【作者】 谢明辉;
【导师】 张嗣良;
【作者基本信息】 华东理工大学 , 生物化工, 2013, 博士
【摘要】 目前计算流体力学(CFD)方法已经在生物过程反应器的设计和放大过程得到了应用。但是CFD方法与生物反应过程模型相结合来预测生物反应过程,并实现过程优化与放大是有困难的。反应器内多相流的流动、混合和传递过程是复杂和多维的,然而生物反应器内非牛顿型的发酵体系的流体动力学行为则更为复杂。本文以多层搅拌式生物反应器为对象,开展了单相、以及不同溶液中的气液两相流的实验流体力学与计算流体力学研究。采用角度解析的大涡PIV流场测试技术发现径流RT和HBT桨在桨叶的后方形成一对相反方向的尾涡,而翼型WHHd和WHu桨在桨叶后方形成一个尾涡。并且发现角度解析的最大湍动能的值约为时间平均的最大湍动能的2倍,最大湍动能耗散率约为时间平均的最大湍动能耗散率的3.5倍。实验结果发现湍能耗散率积分值依赖于模型参数Cs,当取Cs=0.12,大涡PIV方法得到的湍动能耗散率体积积分的功率与扭矩的测得的功率便能相等。3RT桨的桨叶区的功率占22.1%,桨叶排出区占45.2%,主流区占32.7%。对于上翻操作的3WHu组合的桨叶区的功率占39.2%,桨叶排出区占23.3%,主流区占37.5%。组合桨中的每层桨都具有自己的特征。RT桨的最大湍动能耗散率与平均的湍动能耗散率的比值(εmax/εavg)的平均值为18.6,HBT桨的εmax/εavg为26.9~34.1,WHd桨的εmax/εavg的平均值为22.8,WHu桨的εmax/εavg的值为14.7~23.1。空气-水溶液两相体系中,在低表观气速时(通气量为0.2vvm)上翻型搅拌器的传质能力优于下压型搅拌器,3WHu的传质系数比HBT+2WHd高53%, HBT+2WHu和3RT的传质能力居中。而表观气速较高时(通气量为1.0vvm),在相同的比功率输入情况下所有搅拌组合的传质能力相近。开发的三电导探针测得了局部的气含率表明:对于3RT,最高的气含率位于底层桨叶排出区,其次是中层和顶层桨排出流的上下方靠近壁面的位置。HBT+2WHd的底层桨的气含率与RT桨相似,主体区气含率分布比较均匀。HBT+2WHu和3WHu组合中,两层上翻型桨叶之间的气含率较高,项层桨上方的气含率相对较低。气泡的速度场分布表明在通气情况下,RT桨同样形成两个循环,但下方的循环比上方的更靠近器壁。在搅拌控制的条件下,混合时间随着通气量的增加而增加。总的来说,上翻型桨叶组合3WHu和HBT+2WHu组合的传质和混合特性是比较好的。采用低浓度的CMC溶液替代菌丝发酵液来研究反应器内气泡大小和传质特性。发现相同功率输入时,上翻操作的桨型组合的气含率要高于下压操作以及3RT组合产生的气含率,气含率εG∝(正比于)(PG/V)0·3VG0.62。相同的通气功率下,Sauter平均直径大小依次是:3Whu>HBT+2WHu>HBT+2WHd>3RT,而相界面积大小排序是:3RT> HBT+2WHd>HBT+2WHu>3Whu,这是因为上翻桨叶气泡的聚并现象比较严重,Sauter平均直径d32∝(PG/V),)-0.12μGa0.32εG0.14,相间面积a∞(PG/V)-0.4VG0.5μa-0.5。低浓度的CMC溶液上翻操作的桨型组合的传质能力最好,但是随着粘度的增加,上翻桨叶组合的传质能力下降较快,气液传质系数kLα∝=(PG/V)0.5VG0.45μα-0.78。WHu、HBT+2WHu的kL是3RT、HBT+2WHd的kL的1.5-2倍,液相氧传质系数kL∝=(PG/V)0.11μα-0.24。气液传质依赖于桨型结构性质,实质是取决于不同桨型结构产生不同的流场,与物性一起决定了气泡的动力学和传质性能。本文研究发现高粘黄原胶溶液要达到相同功率消耗,小直径桨型组合的转速随着浓度的增加而增加,而大直径桨型的转速随着浓度的增加而降低。局部kLα分布发现黄原胶溶液的浓度为1.Owt%时,小直径桨型组合罐内的流体混合较好,kLα分布比较均匀,但随黄原胶浓度增加时壁面基本上变成传递的死区。而大直径桨型组合除了罐底部区域,kLa基本分布比较均匀,但是随着浓度的增加,大直径桨型组合基本丧失了气体分散的能力。实验发现桨型组合的平均kLa值大小排序为3RT> HBT+2WHu> HBT+2WHd> EG> HBT+2MIG> HBT+DHR。且发现功率与表观粘度对kLα影响较大,通气量对kLα的影响较小。小直径桨组合的混合时间要大于大直径桨叶的组合,同时也发现粘度对混合时间的影响要大于功率消耗对混合时间的影响。采用不同桨型组合研究流场特性对黑曲霉产糖化酶的发酵实验的影响。结果发现保证OUR趋势一致的情况下,三种桨型最终的转速3WHu>3RT> HBT+2WHu,对应的功率输入是3RT>3WHu> HBT+2WHu。而最终3RT的菌体酶活是最低的,上翻式桨叶组合HBT+2WHu和3WHu的菌体酶活比较接近。3RT由于高剪切而延迟菌球出现的时间并且菌球的浓度较低,上翻式桨叶组合由于较小的剪切,其菌球出现早而且其浓度也大于平叶桨的菌球浓度。发酵过程的发酵液流变特性的测试表明HBT+2WHu的表观粘度要低于3RT罐内的表观粘度。这也说明菌球的形成确实有利于降低发酵液的黏度,促进反应器的流动性,从而提高了产酶效率。建立了HBT+2WHu桨型组合的发酵过程kLa关联式。这些研究结果为未来的工艺优化及放大提供了有益的线索和理论指导。对于3RT的单相数值模拟,基于雷诺平均的湍流粘性模型计算得到速度值比PIV测试值的要大,湍动能的值都要比实验值低40-80%。而LES得到的流场与PIV测试的非常相似,包括速度与湍动能分布,排出区的湍动能比实验值约低15-30%,主流区的湍动能和PIV测的湍动能吻合很好。对于气液两相流的模拟采用CFD和PBM耦合求解。采用Ishii-Zuber曳力模型模拟形成的气含率较低,改用修正的Brucato曳力模型能获得比较满意的气含率。High Resolution对流差分格式与修正的Brucato曳力模型,增强系数4.5×10-6的模拟方法预测的气含率、相界面积以及kLα与实验值比较接近。kLα的增加是其通过增加曳力来增加相界面积,且认为CFD模拟的kLα值比实验值低的主要原因是低估了罐内的湍动能耗散率,其次低估了曳力。
【Abstract】 Computational fluid dynamics (CFD) method has been used in the bio-process for reactor design and process optimization. Combination CFD method with bio-processes model to predict bio-processes and achieve process optimization is difficult. The flow field, mixing and mass transfer of multiphase flow are very complex. The hydrodynamic of the bioreactors are more complex due to the non-Newtonian broths. In this paper, stirred bioreactors with multiple impellers were employed to study the hydrodynamics in single-phase and gas-liquid flow with different media by experimental and CFD methods.It can be found that there are a pair of trailing vortices behind the blade of RT and HBT impellers, while only a trailing vortex behind the blade of WHd and WHu impellers by the angle-resolved large eddy PIV measurement techniques. The maximum turbulent kinetic energy of angle-resolved is about2times of that obtained by the time-averaged, and maximum turbulent energy dissipation is about3.5times of that obtained by time-averaged. When Cs=0.12, the power getted from volume integral of turbulent energy dissipation is equal to the power calculated by the measured torque. For3RT,22.1%of the total input energy is dissipated in the impeller region,45.2%in the stream region and32.7%in the remaining volume of the tank. For3WHu,39.2%of the total input energy is dissipated in the impeller region,23.3%in the stream region and37.5%in the remaining volume of the tank. Each impeller of the combination has own characteristic even the same type. εmax/εavg is the ratio of maximum turbulent energy dissipation to the average turbulent energy dissipation. The average εmax/εavg of RT impellers is18.6, the εmax/εavg of HBT impellers is26.9-34.1, the average εmax/εavg of WHd impellers is22.8, the εmax/εavg of WHu impellers is14.7-23.1.In the air-water system,3WHu produces53%higher mass transfer coefficient than HBT+2WHd, HBT+2WHu and3RT lie between them at gas superficial velocity. At high gas superficial velocity, however, all the tested configurations give almost similar mass transfer coefficient under equivalent power input. For3RT, highest hold-up is in the bottom impeller discharge stream and near the wall for the middle and top impellers. For the HBT+2WHd combination, there was no large variations of gas hold up in the bulk except region around the bottom impeller. For HBT+2WHu and3WHu, high gas hold-up was observed between the two up pumping impellers, and moderately low gas hold-up above the top impeller. Under gassed condition, each RT also generates two loops, but the lower loop is more near the wall than the upper loop. the mixing time increases with the gas flow rate increases in the agitator-dominated regime. Overall, mass transfer and mixing characteristics are quite good for the3WHu and HBT+2WHu combinations.CMC solutions with different concentrations which are preferred to mycelial fermentation broth were used for the study of bubble size and mass transfer characteristics in bioreactor. At same power input, the impeller combinations with up-pumping produce higher gas holdup than that of other combinations, gas holdup εG∝(PG/V)0.3VG0.62. At same power input, the rank of Sauter mean diameter is:3Whu> HBT+2WHu> HBT+2WHd>3RT, and the rank of interfacial area is:3RT> HBT+2WHd> HBT+2WHu>3Whu, this is due to more bubble coalescence for up-pumping impeller. Sauter mean diameter d32∝(PG/V)-0.12μα0.32εG0.14, while interfacial area α∝(PG/V)-0.14VG0.5μα-0.5. At low concentration of CMC solutions, impeller combinations with up-pumping give best mass transfer, the mass transfer coefficient kLα∝(PG/V)0.5VG0.45μα-0.78. Liquid phase mass transfer coefficient kL of WHu and HBT+2WHu is1.5-2times higher than that of3RT and HBT+2WHd, and kL∝(PG/V)0.11μα-0.24. Mass transfer depends on the flow fields gernerated by impellers and physical properties, because they determine the bubble dynamics and mass transfer performance.For highly viscosity xanthan gum solutions, in order to gain the same power input, the rotating speed of "small-diameter" impeller combinations increases as the concentration of xanthan gum increases, while it decrease for "large-diameter" impeller combinations.. For the "small-diameter" impeller combinations, the kLα value near the wall drop faster than other areas as the concentration of xanthan gum increases. While for the "large-diameter" impeller combinations, the kLα distribution is homogenous except the bottom area but with poor gas dispersion capability as concentration of xanthan gum increases. The rank of average kLα is:3RT> HBT+2WHu> HBT+2WHd> EG> HBT+2MIG> HBT+DHR. The obtained correlation shows that the kLα is heavily depend on specific power and viscosity, but less influenced by the gassing rate. The mixing time of small-diameter impeller combinations is greater than that of large-diameter combinations, the effect of viscosity on the mixing time is greater than that of power consumption on the mixing time.The fermentation of Aspergillus niger to produce glucoamylase with different impeller combinations, in order to keep same OUR, the final impeller speed is3WHu>3RT> HBT+2WHu, corresponds to the power input is3RT>3WHu> HBT+2WHu. Ultimately, the emzyme activity of3RT is the lowest, the emzyme activity of HBT+2WHu and3WHu are closer. High shear strain rate of3RT delays the formation of pellets and gains lower pellet concentration. Howerver, low shear strain rate of HBT+2WHu accelerates the formation of pellets and gains higher pellet concentration. The rheological properties of the fermentation broth show that the apparent viscosity in the bioreactor with HBT+2WHu is lower than that in the bioreactor with3RT. This also show that the formation of pellets do help to reduce the viscosity of the fermentation broth, and promote the mobility of the reactor, thereby increase the production efficiency of enzyme. Impellers with up-pumping have excellent mixing and low shear rate for the fermentation of filamentous fungi, and the kLα correlation for fermentation process have established for HBT+2WHu. These results provide useful clues and theoretical guidance for future process optimization.For the single-phase numerical simulation of3RT, the speeds obtained from Reynolds-averaged turbulence model are larger than the PIV data, but turbulent kinetic energy values are40-80%lower than the experimental values. The flow field getted by LES simulation are very similar with the PIV results, the values of turbulent energy dissipation are approximately15-30%lower than that of the experimental values in the stream region. The values of turbulent kinetic energy in other regions agree well with PIV results. CFD coupled PBM model was employed for numerical simulation of multi-phase flow. Simulated gas holdup is low using Ishii-Zuber drag force model. The modified Brucato drag force model can obtain satisfactory gas holdup. The simulation method of Hi Resolution convection difference scheme combining modified Brucato drag force model, enhancement factor4.5×10-6can predict well the gas holdup, interfacial area and kLα. the kLα value increases is due to the increase of the interfacial area results from the increase of drag force. The simulated kLα value is ower than the experimental value is mainly due to underestimation of the turbulent energy dissipation in the tank, followed by underestimation of the drag force.
【Key words】 stirred bioreactor; mass transfer; mixing; CFD simulation; large eddysimulation; PBM model;