1

Bottom portion (no solids)

Grid

Cylindrical pipe . with dish end and bottom orifice

FIGURE 9.7 Schematic of mixing cup (from Ranade, 1999b).

of air to these elements will lead to mal-distribution of oxygen in the fluidized bed. The layout of these elements on the grid will have an influence on gas distribution within the fluidized bed reactors.

These issues indicate that it is crucial to obtain quantitative information about fluid dynamics and mixing in mixing elements as well as in the overall reactor in order to evaluate the possibility of capacity enhancement using oxygen-enriched air. One of the concerns is to examine the scenario of malfunctioning of a mixing element and to evaluate and compare the consequences for ordinary and enriched air feed. Prior experience and knowledge about the catalyst and kinetics of the reaction suggested that catalyst and process conditions were adequate to convert additional feed, provided the reactor is operated in the same flow regime. Due to this fact and the limited time available, it was decided that it is sufficient to develop a computational model to simulate gas flow and mixing within the industrial fluidized bed reactor without considering any reactions. This decision facilitated rapid development and application of the computational model. Possible consequences must, however, be kept in mind when interpreting and applying the results obtained from the computational flow model. When developing a computational flow model for OXY reactors, several demands imposed by the set objectives should be kept in mind. It is necessary to consider the overall configuration of OXY reactor to evaluate possible mal-distributions. The construction of an industrial OXY reactor is extremely complex. It is normally fitted with internal cooling coils to remove and recover the heat of reactions. The presence of internal cooling coils may also lead to mal-distribution. The bottom conical portion, grid containing mixing elements, feed distributors, internal supports, etc. makes the task of modeling the geometry of an industrial reactor quite complex. To restrict the computational demands, an appropriate modeling methodology needs to be evolved. Here we discuss the methodology used by Ranade (1999b) to simulate flow in a complex industrial OXY reactor by developing different modeling layers and some of his results. The methodology will be useful for identifying and enhancing the capacity limits of existing reactor hardware.

9.3.2. Methodology to Simulate Flow in an OXY Reactor

Construction of a large-scale industrial OXY reactor is extremely complex. A wide range of length scales, ranging from a few meters (reactor diameter and length) to a few millimeters (orifice of mixing element), appears to be important. There is a corresponding two to three orders of magnitude difference between gas velocities in the various elements of an OXY reactor. Resolving these widely different length and velocity scales (typically in the range, 0.01 to 4 m and 0.2 to 100 m s-1) simultaneously may stretch the limits of computational resources. It is, therefore, necessary to evolve a suitable modeling strategy. It must be emphasized that at every stage, the reactor engineer must be aware of the underlying assumptions (explicit and implicit) and their consequences, when interpreting and using the results.

The overall problem of modeling the fluid dynamics of an industrial OXY reactor was first divided into several small components (mixing element, bottom conical portion of the reactor, portion above the grid). Each of these components was studied using a separate model. The understanding gained through these studies was then combined to construct the model for the whole reactor (the methodology is shown pictographically in Fig. 9.8). From the discussion of Section 9.3.1, it is clear that it is essential to resolve all the fine-scale flow characteristics of the mixing element. It may not be necessary (or possible) to resolve such fine scales when simulating flow in the bottom portion or upper portion of the OXY reactor. Therefore, in the present work, a three-layer modeling strategy was used. A computational model was first developed to understand the fluid dynamics of the mixing elements. Apart from the mixing, it was also used to characterize the pressure drop across the mixing elements under different operating conditions. In the second layer, gas flow in the bottom portion of the reactor was modeled to examine possible mal-distribution of feed air. Flow in the top portion above the grid was modeled to examine the influence of layout of mixing elements on the grid. In the third layer, gas flow in the complete reactor (excluding the internal cyclones) was simulated. The grid plate supporting the fluidized bed was modeled as a porous plate with pressure drop characteristics obtained from results of the first modeling layer. An attempt was made to evaluate and synthesize information obtained from all three modeling layers, validate these whenever possible (either directly or indirectly) and then use the information to evaluate various scenarios.

Suitable computational models for each of the layers discussed above were developed on the basis of available information and a time scale analysis of flow in OXY reactors (see Ranade, 1999b for more details). Because of the magnitude of pressure drop across the grid, it was found necessary to employ compressible flow equations. An ideal gas assumption was used to calculate the density of gas at any point (as a

FIGURE 9.8 Modeling methodology (from Ranade, 1999b).

function of local pressure, temperature and effective molecular weight of the gas). Under normal operating conditions, the flow was found to be turbulent. Turbulent stresses were modeled using a suitable two-equation turbulence model (see Chapter 3). In order to capture the influence of the shape of the mixing cup, it was modeled using a body-fitted grid. Suitable boundary conditions were developed. The bottom part of the OXY reactor containing the bottom conical portion, the grid and the top portion up to the cooling coils was modeled using body fitted grids. Air feed was modeled by prescribing appropriate mass and momentum sources to the cells located just below the air feed pipe. The open area of the grid plate was modeled as a porous media. Appropriate sources of ethylene and HCl were specified at grid porous cells to simulate feed pipes. The characteristic resistance of these porous media was prescribed from results obtained for a single mixing element. The coil bundle inside the reactor was modeled as porous media and the top surface of the solution domain was modeled as a constant pressure surface. Six species namely, oxygen, nitrogen, inert, water, ethylene and HCl were considered for these simulations. The computational model was mapped on to a commercial CFD code, FLUENT (Fluent Inc., USA) with the help of user-defined subroutines.

9.3.3. Fluid Dynamics of OXY Reactor

(a) Mixing element: Flow in a mixing element (cup) is governed mainly by the pressure drop across the grid plate. Pressure outside a mixing cup is higher than that prevailing over the annular open space of the cup. This pressure difference causes airflow from the surrounding space into the cup through the bottom orifice. This high velocity air jet impinges on the jet of ethylene and HCl mixture, generating intense turbulence and leading to complete mixing of the two streams, which escape from the annular opening of the cup.

Simulations were first carried out to examine the relationship between pressure drop across the grid and the resulting airflow through the cup. For all these simulations, the flow rates of ethylene and HCl were set to pre-determined values corresponding to reactor loading. The pressure drop across the cup was then adjusted to get the right amount of airflow through the cup. These simulations were carried out for three reactor loadings. After verifying the model predictions, by comparing these with plant data (see Fig. 9.9), the computational model was used to gain an insight into the fluid dynamics of the mixing elements. Predicted contours of stream function in the mixing cup are shown in Fig. 9.10a. It can be seen that the two impinging jets (air jet from the bottom orifice and mixture of ethylene and HCl from the pipe) generate a complex re-circulating flow within the mixing cup. The incoming, high velocity jet of air through the bottom orifice causes formation of two re-circulating loops near the walls of the mixing cup. There is relatively little exchange between these recirculating loops and the high velocity upward flow. Predicted contours of oxygen mass fraction and ethylene mass fraction within the mixing cup are shown in Fig. 9.10b.

The stagnation point formed by the two impinging jets is located near the end of ethylene and HCl feed pipe. Particle streak lines calculated based on the predicted flow results (not shown in the figure) clearly indicate the location of the stagnation point,

0 0

Post a comment