Many of the probes used to understand hydrodynamics in circulating fluidized bed risers intrude into the environment they are measuring, although assumptions are typically asserted that the intrusive probes do not affect the data collected. This could be a poor assumption in some cases and conditions. We found that intrusive fiber‐optic probe measurements consistently mis‐predicted the solids concentration compared to the nonintrusive pressure drop measurements outside the fully developed flow region of a riser containing fluid catalytic cracking catalyst or glass bead particles. The discrepancy was sensitive to superficial gas velocity, solid circulation rate, probe position, and flow direction. Barracuda VR™ computational fluid dynamics simulations confirmed this, and indicated that particle momentum was lost at the leading edge of the probe and particles were spilling over to the probe tip. Accordingly, new probe designs were proposed to mitigate the intrusiveness of a fiber‐optic probe for more accurate characterization. © 2017 American Institute of Chemical Engineers AIChE J , 63: 5361–5374, 2017

Pressure taps are ubiquitous in fluidized bed reactors. Unfortunately, attention is typically paid only to the mean or standard deviation of the fluctuations in practical operations, which leads to a loss in the wealth of information contained therein. This effort is targeted at assessing the hydrodynamic characteristics that can be understood from a comprehensive data set of pressure signals via wavelet decomposition. Differential pressure (Delta P) signals were recorded at 11 axial positions along a pilot-scale CFB riser for five particle systems (namely, three monodisperse materials and two binary mixtures) and four operating conditions. The effects of material property, operating condition, and riser axial position on the differential pressure (Delta P) signals were evaluated by comparing the energy contained in the wavelet-decomposed signals (E-Dj). Results indicate the following: (i) whereas the averaged Delta P values and power spectra of the signals were largely invariant, E-Dj values tended to be sensitive to material property and operating conditions throughout the riser; (ii) under conditions of lower solid loading, higher E-Dj values were obtained and E-Dj was more sensitive to material properties; (iii) among monodisperse particles, the larger and denser particles tended to be most sensitive to operating conditions; (iv) among the five particle systems, monodisperse small glass and monodisperse large glass systems gave, respectively, the lowest and highest E-Dj values; (v) particle density exerts a more dominant influence than particle diameter on E-Dj; and (vi) the effect of axial position on E-Dj was secondary compared to that of operating condition and material property. The observations provide insights into Delta P signals from CFB risers, which find value spanning a mechanistic understanding of the relationship between the gas phase and particles and process monitoring in practical operations.

Solids flow direction near the wall of a CFB riser was studied for three monodisperse, two binary mixtures and one continuous particle size distribution (PSD) of Geldart Group B particles. Measurements were taken at five axial positions along a 0.3m diameter, 18m tall pilot-scale CFB riser at four operating conditions. The data were compared to predictions available from flow regime maps. Results showed that (i) for monodisperse systems, particle diameter was more dominant in dictating upflow annulus behavior; (ii) the binary mixtures exhibited largely upflow annulus behavior, presumably due to the dominant influence of the larger constituent particles in dictating upflow annulus behavior; (iii) for the continuous particle size distribution (PSD), the impact of this type of polydispersity was more pronounced at lower solid loading conditions; and (iv) the applicability of the available regime maps, which were largely developed based on Geldart Group A particles, was found to be limited for the Geldart Group B and non-monodisperse particle systems investigated in this study.

Particle clusters are well acknowledged to affect the hydrodynamics and overall performance of gas-solid fluidized beds. Since one of the first reports on the clustering phenomenon in 1948, the understanding of particle clustering has been rigorously attempted via both modeling and experimental efforts, with significant traction gained especially in the last few decades. Accordingly, the current review targets at providing a comprehensive landscape of the experimental cluster trends to summarize the findings, in particular on circulating fluidized bed (CFB) risers, to date. More questions than answers seem to have sprouted from the abundant experimental data available, which impedes model development. The quantitative comparison of cluster characteristics across studies must be treated with caution, because of (i) different riser configurations, instruments and analysis methods, which can lead to discrepancies of an order-of-magnitude; (ii) the impact on cluster characteristics by an interplay of a host of factors, hence the influence of a single parameter is not straightforward, even within the same study; (iii) the irregularity in the form of clusters, hence the definition and/or measurement of the various cluster characteristics differ; and (iv) the general lack in the reporting of the actual particle size distribution. What is remarkable is that the trends of the cluster characteristics are relatively consistent despite different experimentalists and units.

Experiments involving monodisperse Geldart Group B particles have been carried out in a pilot-scale riser of a circulating fluidized bed (CFB). Several combinations of superficial gas velocity (Us), solid flux (Gs), average particle diameter (dave), and particle material density (ρs) were investigated. Surprisingly, the experiments reveal the presence of a reverse core-annulus profile (i.e., a dense core with a dilute annulus) under certain conditions. Specifically, for the large glass beads (dave=650μm, ρs=2500kg/m3), the reverse core-annulus profile was observed near the top of the riser for all Us and Gs combinations examined. For high-density polyethylene (HDPE) pellets (dave=650μm, ρs=900kg/m3) of the same dave, reverse core-annulus was observed at the top of the riser only at relatively low Gs. However, for the smaller glass beads (dave=170μm, ρs=2500kg/m3), the traditional core-annulus profile was observed for all Us and Gs combinations. Although previous work provides possible explanations for this behavior (gas-phase turbulence, etc.), the evidence obtained in this system suggests a novel dominant factor for reverse core-annulus flow: the particle Stokes number (St). Lower-St particles are more apt to follow the gas exiting the riser, while higher-St particles have a longer relaxation time and thus are more likely to re-enter the riser after collision with the roughened rounded-elbow exit. Accordingly, the re-direction of particles from the rounded-elbow exit and back into riser due to large-scale roughness along the elbow is greater for higher-St particles.

In a FLUID COKING unit, reactor cyclone fouling by coke deposits can set the run length of the unit. Over time the coke deposits can grow and obstruct the cyclone which will limit throughput and lead to a shutdown. For this reason, producing a more uniform coke distribution pattern within the reactor horn chamber may lead to an increased interval between turnarounds. An existing pilot-scale experimental model of the coker reactor freeboard, horn chamber, and exit cyclones allows determination of coke distribution to the cyclones, but provides limited understanding of the Underlying fluid dynamics within the system. In this work a two-phase computational fluid dynamics (CFD) model of this experimental rig was developed. Coke. was modeled as an Eulerian stream of solid particles with monodisperse particle diameter. It was found that predicted coke distributions were sensitive to the choice of coke diameter, but a suitable choice gave good agreement with experimental observations. In the current work this value was 167 mu m which was substantially higher than the value of the Sauter mean diameter of 139 mu m. It was found that the CFD model could quantitatively predict coke distributions in the freeboard region of a FLUID COKING reactor experimental rig, while providing insight into the flow dynamics. When modeling the particle size distribution with a monodisperse particle diameter, comparison with experimental results is necessary to identify the coke particle diameter that leads to optimal model performance.

Experiments directed at understanding local mass flux behavior of Geldart Group B materials in the riser of a gas-solids circulating fluidized bed (CFB) have been carried out. Three monodisperse materials (with differences in particle size and/or material density), two binary mixtures (one with only a particle size difference between the species and the other with only a material density difference), and one continuous particle size distribution (PSD) have been investigated at four operating conditions. Results show that the riser axial position has the greatest influence on mass flux behavior, especially near the top of the riser, where profile shapes consistently have an inverted U-shape or V-shape. The material type (i.e., monodisperse materials of different particle sizes and/or particle densities or different types of polydispersity) and operating conditions effects are secondary but more apparent at the riser bottom. An interesting observation involving binary mixtures is that while the mass flux profiles of the density-difference binary mixture mimics that of one of its (monodisperse) constituent components, the size-difference binary mimics neither of its two monodisperse components.

Experiments in a circulating fluidized bed (CFB) riser with Geldart Group B particles have been carried out with an emphasis on cluster characterization. In this paper, the focus is on monodisperse materials; results for polydisperse materials are contained in the companion paper (Chew et al., this issue). A fiber optic probe was used for detection of solids, with the resulting solids cluster identification accomplished using wavelet decomposition. Three cluster characteristics were evaluated – appearance probability, duration, and frequency – as a function of riser position, operating condition, and type of solids (i.e., variations in particle size and/or material density). Results indicate that the riser position has the most dominant influence on cluster characteristics with the cluster appearance probability appearing to be largely a function of riser position only, and insensitive to changes in operating condition or material type. The cluster duration and frequency are influenced by the particle material and operating condition, though only from mid-height of the riser upwards. The results presented provide a comprehensive picture of factors affecting cluster trends.