035

Height from the bottom, m

A: experiment; A: fine mesh; □: medium mesh; o: coarse mesh; ■: granular model

FIGURE 12.21 Comparison of simulated (constant viscosity or granular model) and experimentally observed bubble frequency (Pressure 0.8 Mpa, from Enwald et al., 1999).

develop a comprehensive CFD model to simulate an entire reactor operation. Insight gained through application of CFD models may, however, lend significant help in engineering decision making. Two examples of the application of CFD-based models to enhance the performance of fluidized bed reactors are discussed in Chapter 9. Here we illustrate the potential by describing more applications from published literature.

Use of CFD models to understand mixing and mal-distribution issues in an oxy-chlorination reactor was discussed in Chapter 9. Samuelsberg (1994) used a two-fluid model with kinetic theory of granular flows to simulate a bubbling fluidized bed carrying out oxi-chlorination reactions. They simulated only a two-dimensional slice of the oxi-chlorination reactor. The internal heat transfer tubes were modeled as a porous block (volume blockage 0.75) assuming a constant bed-to-wall heat transfer coefficient. The simulated spatial distribution of ethylene di-chloride (EDC) is shown in Fig. 12.22. The highly dynamic behavior of bubbling beds is clearly evident from this figure. It can also be seen that the most significant EDC formation takes place in the bottom part of the reactor. Though this model employs a strongly simplified representation of a full-scale oxy-chlorination reactor and internal heat exchanger, it provides much more detailed information than conventional design and development models. The possibility of hot spot formation and interaction of gas mal-distribution on reactor performance can be understood using such a computational model. A discrete particle model was used by Kaneko et al. (1999) to understand similar issues in a bubbling polymerization reactor. This model was used to understand differences between porous and perforated plate gas distributors in terms of reactor performance and hot-spot formation. A sample of their results is shown in Fig. 12.23. It can be seen that simulations indicate the possibility of formation of a dead zone for the perforated distributor. The temperature of gas in such a dead zone may increase significantly during the course of the reaction.

360s2 380s2 400s2 420s2

360s2 380s2 400s2 420s2

11)11 ] i i t i tir>« «tili 440s2 460s2 480s2 „ 500s2

FIGURE 12.22 Simulated spatial distribution of ethylene di-chloride (from Samuelsberg, 1994).

251iniformiontours2

black<0.2

white^0.472

FIGURE 12.22 Simulated spatial distribution of ethylene di-chloride (from Samuelsberg, 1994).

CFD-based models have also been used to simulate the performance of riser reactors (Theologos and Markatos, 1993; Theologos et al., 1997; Gao et al., 1999). Gao et al. (1999) developed a three-dimensional two-phase turbulent flow model to simulate a FCC riser reactor. A thirteen-lump kinetic model was used to simulate cracking reactions taking place in the FCC riser. The computational flow model was first verified by comparing predicted results with the published experimental data of Bader et al. (1988) and Yang (1991). A sample of their results is shown in Fig. 12.24. It can be seen that their computational model was able to capture key flow characteristics of gas and solid phase quite adequately. This is crucial for extending the use of computational flow models for reactor engineering. Gao et al. (1999) then used this computational model to simulate the industrial riser reactor. In such a reactor, feed is usually introduced at the bottom of the riser via multiple nozzles. Conventional one-dimensional riser models cannot account for radial variation of catalyst and feed concentrations. The CFD-based three-dimensional model of Gao et al. (1999) could capture such radial as well as axial variations. Local fluid dynamics near feed nozzles

Porous plate

Perforated plate (5 holes)

0 0

Post a comment