Data from experiments on bubbling ﬂuidization of group B glass particles

Experiments on bubbling ﬂuidization were performed in three cylindrical columns having internal diameters of 2.5, 4 and 6 inches. Though not scaled hydrodynamically, the systems are designed to have considerably high particle count compared to major-ity of controlled experiments reported in the literature. A systematic testing procedure was followed involving replicates and randomization to estimate uncertainty and avoid unintended bias. Glass particles having a sauter mean diameter of 332 µ m were used and the superﬁcial velocity of air at the inlet was varied from 2.97 - 5.35 times minimum ﬂuidization velocity. Mean and standard deviation of diﬀerential pressure and interface height are the quantities of interest reported in this work. The results obtained from these experiments are found to be consistent with the previous studies. Besides eluci-dating the underlying physics, such datasets are critical to assess predictive capability of coarse-grained modeling techniques like Particle-In-Cell (PIC) or Coarse-Grained Discrete Element Model (DEM) developed for large-scale applications .


Introduction
Fluidization has been widely used in the field of chemical engineering due to favorable mixing and heat and mass transfer characteristics [1][2][3][4][5][6] . Several researchers including Glicksman 7 , Horio et al. 8 , van den Bleek and Schouten 9 , Schouten et al. 10 , Briongos and Guardiola 11 have analyzed this process to aid in design and scale-up of efficient reactors. For instance Glicksman 7 proposed several non-dimensional groups to be considered while scaling including , ρg ρs , L dp , D dp which represent Froude number, modified Archimedes number, density ratio, ratio of bed height to particle diameter, and ratio of bed diameter to particle diameter respectively. Albeit, it might not be possible to consider all the non-dimensional groups at once to scale a system from laboratory-scale to pilot-scale or industry-scale. This would require changing the properties of material as well as fluidizing medium which might not be practical 12 .
It is more feasible to scale the geometry while the properties of material and fluidizing medium are held constant. 8,[13][14][15][16][17] Even though this approach might not be consistent with a true scale-up effort, these experiments are essential to enhance our understanding of multi-phase dynamics at different scales. This is vital from a Simulation Based Engineering (SBE) point of view, which aims at predicting the behavior at different scales using highquality experimental data or results from numerical simulations having a high resolution.
The work presented contributes to the existing repository of experimental datasets related to the Quality Assurance program of the Multiphase Flow Science group at National Energy Technology Laboratory which develops and maintains the open-source software MFiX (Multiphase Flow with Interphase Exchanges). Over the past few years, there has been an increasing emphasis on Verification, Validation and Uncertainty Quantification applied to granular and multiphase flows [18][19][20] . This is made possible by active collaboration between experimental and computational groups as envisioned by the ASME V&V (Verification and Validation) guidelines 21 .
Experiments have been designed in the current study to obtain objectively assessed data which could be used for validating coarse-grained techniques such as Particle-In-Cell (PIC) or Coarse-Grained Discrete Element Model (DEM). Computations using conventional Langrangian-based strategies, for example DEM, pose a limitation while modeling industrialscale systems where particle counts become intractable. On the other hand, Eulerian Two-Fluid Model (TFM) is prone to inaccuracies while modeling static or near-static regions where the motion of solids is mainly governed by inter-particle friction 20,22 . Also, adding multiple components to a TFM framework introduces greater uncertainty in constitutive relations 23 . Coarse-grained techniques have thus been increasingly favored, however highquality data for benchmarking such methodologies are very scarce in literature.
In this analysis, three geometries having internal diameters of 2.5 inch, 4 inch and 6 inch were considered. Glass beads having a Sauter mean diameter of 332 µm were used in these experiments. Flow velocity of air at inlet was varied from 2.97-5.35 times the minimum fluidization velocity, while the range of particle Reynolds number and density ratio were held constant across the experiments. Measurements of differential pressure in different regions and interface height are reported. Results obtained from this systematic study provide a valuable database for a rigorous validation and uncertainty quantification of computational models and constitutive relations besides aiding in understanding the hydrodynamics at different scales.

Experimental Setup and Procedure Preliminaries
Glass particles used in this study were characterized using QICPIC (manufactured by Sympatec GmBH). Distribution density of particle diameter is shown in Figure 1. These particles have a sauter mean diameter of 332 µm and are classified under Group B 24 characterized by good mixing behavior and vigorous bubbling 2 . Figure 2 shows the cumulative distribution of sphericity and aspect ratio, having a median value of 0.93 and 0.97 respectively. In addition, Figure 1 also shows an example of distribution density from the 4-inch cylindrical column after fluidization experiments thereby confirming negligible effects due to fragmentation or attrition. It is important to identify changes in size distribution to make the validation study more consistent.
Following particle characterization, tests were performed to estimate minimum fluidization velocity, U mf in all the columns. Further details regarding the procedure can be found in Vaidheeswaran at al. 25 . Table 1 30 suggest that the friction effects due to wall increases as the hydraulic diameter is reduced which leads to an increase in U mf for smaller geometries. Hence, a greater gas-phase inertia is required to support the weight of bed material and overcome resistance due to wall.

Fluidization Experiments
Fluidization experiments were performed in cylindrical test sections having internal diameters 2.5-inch, 4-inch and 6-inch as depicted in Figure 3. The following differential pressure measurements are reported in the current study: ∆P 2 = P 1 − P 2 , ∆P 3 = P 2 − P 3 , ∆P 4 = P 3 − P 6 and ∆P 5 = P 1 − P 6 for the 2.5-inch test section and ∆P 2 = P 1 − P 3 , ∆P 3 = P 3 − P 4 and ∆P 6 = P 1 − P 5 for the 4-inch and 6-inch units. The exact location of pressure ports are summarized in Table 2. Differential pressure signals were recorded at 100 hz for a duration of 180 seconds. High-Efficiency Particulate Air (HEPA) filters were used at the exit to trap particles being elutriated from the units. Tests were performed in a randomized order which includes replicates as shown in Table 3.

Pressure Statistics
Experiments were performed in the order listed in Table 3 in all three columns. Statistics of differential pressure across regions having significant mass of glass particles are reported. section where a slight reduction is noticed. ∆P 3 and σ ∆P 3 are greater than ∆P 2 and σ ∆P 2 in the 4-inch column while they are lesser in the 6-inch column. Also, σ ∆P 2 and σ ∆P 6 measured in 4-inch and 6-inch columns are nearly identical (within measurement uncertainty), where ∆P 6 in these units represents solids in the bed barring the region between distributor plate and the first port above it. It is worth emphasizing that mean and standard deviation of differential pressure are strongly dependent on the location of probes besides operating conditions.
According to Bi 36 fluctuations in pressure signal arise from multiple sources including: (i) forming, coalescence and breaking of bubbles, (ii) bubble eruption at the interface, (iii) passage of bubbles (iv) interactions between fluidized particles and (v) gas-phase oscillations in plenum chamber. Pressure fluctuations due to gas-phase turbulence in plenum occur for distributors of low resistance while those due to particle interactions are dominant very close to the distributor plate 36 . In the current study, we may conclude that most of the oscillations arise due to factors (i), (ii) and (iii) listed above. Furthermore, pressure signals tend to get modulated depending on the dynamics in the fluidizing medium. They get amplified by selfexcited particle oscillations when the bed is sufficiently fluidized or attenuated due to excess energy dissipation. It is hence difficult to characterize the interactions between fluctuations from different sources and associate them with the statistics of differential pressure.

Bed Height Statistics
Bed height or interface location is the second quantity of interest considered in this study.
It represents the boundary between dense bed and freeboard regions. Bubbles being formed closer to the distributor plate propagate upward and erupt once they reach the interface.
This ejects particles on to the freeboard region. In this study, the units are operated in bubbling fluidization regime and hence elutriation of particles is minimized significantly if not eliminated.

Conclusions
The effect of varying U/U mf on bubbling fluidization was investigated using detailed measurements from controlled experiments having well-characterized operating conditions. Mean and standard deviation of the quantities of interest are provided with uncertainty estimates in the form of confidence intervals for differential pressure and threshold dependence for    any of their employees, nor LRST, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.