954 CFD Model

In order to understand the macroscopic circulation within the dense bed, it is necessary to develop a CFD-based model. Such a model will also be necessary to simulate radial non-uniformity in oxygen and temperature distributions. The validated CFD model can be used to evaluate various configurations of air distributor and spent catalyst distributor. As mentioned earlier, it was not possible to use the kinetic theory of granular flows to simulate flow and reactions in the dense bed of a regenerator. Ranade (1998) alternatively used the analogy of bubbles in fluidized beds with bubbles in viscous liquids to simulate macroscopic flow patterns in a bubbling FCC regenerator. The flow in the dense bed of the FCC regenerator was modeled as a two-phase flow comprising an emulsion phase (representing gas-solid mixture with minimum fluidization voidage) and a void phase (representing almost solids-free gas regions within the bed). The two-fluid model of Ranade (1997b) was used for this purpose. The available design and operating information, published correlations and results of the first two modeling layers were used to estimate various relevant scales such as bubble diameter, entrainment, catalyst circulation rate and so on. A wide range of space and velocity scales co-exists in a FCC regenerator. The diameter of the regenerator (~ 5 m) is larger by a few orders of magnitude than the holes of a gas distributor or cyclone standpipes (a few centimeters). It is therefore impractical to resolve all these scales in a single computational model. Appropriate sub-models representing the small-scale internals were developed to make the problem computationally tractable. The results obtained using other modeling layers (kinetic and transport parameters, effective bubble size within the dense bed and so on) were used as input parameters to the CFD model.

The detailed model equations were described by Ranade (1998). In applying two-fluid theory to the bubbling dense bed of a regenerator, it must be noted that the continuous phase density needs to be calculated from the voidage in the dense phase (emulsion phase) and from the density of the catalyst particles. The molecular viscosity of the emulsion phase was specified to be 1 Pa.s based on empirical evidence. The average bubble diameter was specified based on the results obtained using BuDY. The reactions occurring in the dense bed of the FCC regenerator were simulated using species conservation equations (and enthalpy) for the emulsion and bubble phase. Only the principal coke burning reaction (to form carbon dioxide) was considered. Terms representing heat generated due to the combustion of coke and heat transfer from bubble to emulsion phase were adequately modeled. As mentioned earlier, all parameters appearing in the above set of equations were estimated using available data and the results of the first two modeling layers. The governing equations described above were solved using a commercial CFD code, FLUENT (Fluent Inc., USA). Details of the mapping of these model equations onto FLUENT and solution strategies are discussed in Ranade (1998).

The model was used to simulate the macroscopic flow and reactions occurring in a FCC regenerator. A typical grid and predicted velocity field for emulsion phase are shown in Fig. 9.27a and 9.27b, respectively. It can be seen that, except in the region near the regenerator walls, the emulsion phase flows in an upward direction. The predicted velocities in the horizontal planes are much smaller than those in the vertical plane. Fig. 9.27c shows void phase distribution at a typical vertical plane in the dense bed. These flow results were used to simulate coke burning in the dense bed of the regenerator. As expected, higher values of oxygen mass fractions are predicted near the air distributor. The oxygen mass fraction in the void phase quickly drops down as one goes away from the distributor. The knowledge of oxygen breakthrough from the top surface (Fig. 9.28) of the dense bed is important for estimating the extent of after-burning and possible locations of hotspots. The results presented here constitute only a small fraction of the information obtained from the model. The simulation results were analyzed in detail to extract useful information about the behavior of the FCC regenerator. The predicted results were compared with proprietary, plant and experimental data. The predicted extent of solids mixing was found to be less than that indicated by the data. With appropriate tuning of model parameters, however, adequate agreement was obtained between predicted results and available data. The tuned CFD-based model was then used to aid understanding of the macroscopic flow and its influence on reactions occurring in the dense bed of a FCC regenerator. Some results obtained using the model are discussed below.

It is important to develop the capability to simulate the influence of changes in the air distributor on regenerator performance. During the operating life of a regenerator,

FIGURE 9.27 Computational grid and typical predicted results for the FCC regenerator (from Ranade, 1998). (a) Grid, (b) vector plot, (c) contours of gas volume fraction. (Light: high values; dark: low values; legend not shown due to confidentiality constraints.)

FIGURE 9.28 Oxygen breakthrough from the top surface of the dense bed (contours of oxygen mass fraction) (from Ranade, 1998). (Light: high values; dark: low values; legend not shown due to confidentiality constraints.)

the distribution of sparger orifices may change several times owing to either mechanical or operational considerations. The bubble-bubble interaction model (BuDY) was used to simulate the influence of distributor design on bubble formation, coalescence and effective bubble sizes. The CFD-based model was used to examine the influence of distributor design on time-averaged void distribution and macroscopic circulation within the dense bed. The results were useful in understanding the sensitivity of the generated flow to changes in distributor configuration. Examination of the simulation results for the emulsion phase indicates the possibility of short-circuiting spent catalyst particles. To avoid this, several techniques are used in practice, one of the simplest being to install a ski jump type platform near the entry of the spent catalyst. Such a ski jump throws the spent catalyst particles inside the dense bed (instead of releasing them at the regenerator wall), such that the particles enter the upward flowing region of the dense bed. The CFD model was used to evaluate various configurations of spent catalyst entry. Although these results are not included here, it is sufficient to state that the CFD model, along with the other two models (MoBB and BuDY), was used successfully to characterize quantitatively the performance of the industrial FCC regenerator. Results obtained using these models proved quite useful when making engineering decisions regarding gas distributor and spent catalyst entry configurations.

0 0

Post a comment