Computational Investigation of Liquid Maldistribution in Periodically Operated Structured Packed Beds

The non-linear hydrodynamics combined with multiscale transport processes engenders the modeling of packed bed reactors (PBR) a challenging task, particularly in addressing the scale up issues for its commercial application. Familiarized with the steady state hydrodynamics over the past two decades, researchers are now extensively exploring the efficacy of unsteady state operation of packed bed reactors in process intensification and prolong reactor service life. Despite the proven advantages, implementation of periodic flow operation in commercial PBRs is still far from reality, mainly due to process control safety. Although, the application of steady state PBR for commercial purposes is well established, the design concept of periodic PBR needs more attention, even heuristically, to understand the influence of all key parameters. With enhanced capabilities of CFD methods, computational studies can not only complement various facets of physical phenomena but also can help in understanding certain aspects that are generally not attainable by experiments. Surprisingly, limited researches have attempted to develop CFD based models for unsteady PBRs to understand the influence of operating parameters. Moreover, none of these studies have addressed the effect of packing arrangements on flow distribution during periodic operation. Additionally, hydrodynamics of periodic PBRs, fed with non-Newtonian liquids, has rarely been addressed. Cyclic or periodic operation, dictates the periodic toggling of a fluid phase at the inlet, between a low-level (base) and a high-level (peak), typically referred to as the min-max flow operation. This further is termed as the on-off flow when the base flow rate is set to zero.

In this article, we aim to develop a 3-D CFD model for periodic PBRs to understand the effect of particle arrangements and various splits under min-max and on-off operation on flow distribution. A CFD model is developed based on unit cell approach incorporating single phase periodic flow to understand the intrinsic behavior of flow modulation in structured packings. To elucidate the effect of non-Newtonian liquid, a basic power-law model is employed in our validated CFD model. To examine the effect of particle arrangements, face centered cubic (FCC) and modified simple cubic (SC) packing orientations are investigated as shown in Fig. 1.

Figure 1: Schematic showing grids and boundary conditions of (a) FCC, and (b) modified SC arrangements.

Moreover, three different combinations of min-max and on-off operation are chosen under fast mode to identify the influence of flow splits. Liquid distribution is quantitatively examined at the mid-plane of the geometries by dividing the plane into twelve area segments and then evaluating the mass flux through each of the segments. An improved distribution is observed for periodic flow as compared to continuous operation, which is even more pronounced for min-max operation. A distribution coefficient, calculated using the standard deviation concept, indicates that the FCC packing shows the best distribution for min-max operation with the split having maximum peak flow duration as shown in Fig. 2. However, exactly opposite trend is observed in modified SC, where the split with the lowest peak flow rate duration shows the best distribution among the three splits which was even more pronounced for the min-max operation. FCC exhibits better liquid distribution than the modified SC although having a higher pressure drop value due to its lower porosity. Consequently, a comparitive analysis was conducted to determine the desired mode of periodic operation with optimum split ratio that results in least maldistribution. This study will not only find relevance to several packed bed applications but also indicate the periodic flow operation as a viable process intensification technique.

Figure 2: Mass flux distribution at the mid-plane of FCC for different n values at t=10 s for (a) n=0.6, (b) n=1.0, and (c) n=1.2.