Removal of Sulfur Dioxide from a Gas Stream in a Fluidized Bed Reactor Robert J. Best and John G. Yates" Department of Chemical Engineering, University College London, London WC 1E 7JE, England
An experimental study is described in which low concentrations of sulfur dioxide are removed from an air stream at ca. 400 O C by reaction with cupric oxide supported on alumina on a 10-cm diameter tluidized bed. A model of a noncatalytic gas-solid reaction in a fluidized bed is developed and applied to the experimental results and the two are shown to be in good agreement over the range of variables investigated. The implications for the design of large-scale fluid bed desulfurisation reactors are discussed.
Introduction T h e removal of low concentrations of sulfur dioxide from a gas stream such as that emitted from a fossil fuel fired boiler is an industrial problem of widespread concern and is one that may be approached in a variety of ways. I t is clearly a n a d vantage t o retain as much of the sensible heat of the gas as possible in order to assist dispersal from the stack and to this end dry methods of sulfur removal are preverable to those in which the flue gas is water scrubbed. On the other hand, dry methods frequently suffer from all the problems associated with the large-scale handling of solid absorbents and from the fact that capital equipment costs can be high. Nevertheless, the need to reduce the ievel of sulfur emissions is of such general importance that a great deal of work has been done recently to develop new and improved dry methods, and a number of attractive processes have been described. Among these processes is one in which a flue gas a t 350-400 " C is contacted with spheres of alumina impregnated with copper oxide contained in a massive stainless steel cage (Dautzenberg e t al., 1971). In a n oxidizing atmosphere t h e sulfur dioxide present in the gas reacts to produce copper sulfate and the heat of reaction is taken u p by the metal of the cage. Subsequently, the sulfur dioxide may be regenerated a t a higher concentration suitable for sulfuric acid production by reacting the sulfate a t 400 " C with a reducing gas such as hydrogen or a light alkane; in t h e process t h e copper sulfate is converted back to copper oxide. We undertook a study of the sulfur dioxide removal reaction using a laboratory scale fluidized bed reactor since it seemed possible that the process could be developed for operation in a system of coupled fluidized beds with reaction and regeneration proceeding simultaneously and with circulation of solids between the two beds. LVe wish to report here our experimental findings and the comparisons we have made with t h e predictions of a theoretical model of the reactor. These predictions were based on previously reported kinetic data obtained with a packed bed system (Yates and Be:;t, 1976). Experimental Section ( a ) Fluidized Bed Reactor. A schematic diagram of the complete apparatus, which consists for t h e most part of conventional gas handling and metering equipment, is given in Figure 1. The reactor itself was constructed from a stainless steel tube 1.13 m in length, 10 cm inside diameter, a n d with a wall thickness of 2.06 mm. I t was fitted with a Rigidmesh woven stainless steel distributor which was supported on a windbox packed with 1-cm diameter glass beads. Thermocouple tappings were situated a t t h e t o p and bottom of t h e reactor. the lower one being placed immediately above t h e distributor. T h e reactor was heated electrically to a height of
64 cm above the distributor by means of a winding of 23 m of nichrome heating wire dissipating 3 kW; gas supplies to t h e reactor were passed through a preheater which was also electrically heated. Temperature control of both reactor and preheater was maintained by Sirect proportional controllers. T h e whole reactor unit was insulated by glass fiber lagging 5 cm thick. ( b ) Procedure. For a particular series of runs the reactor was operated in a batchwise manner with a single charge of copper oxide-impregnated alumina particles. The preparation and properties of these particles have been discussed in detail previously (Yates and Best, 1976). Their physical properties are summarized in Table I. Reactions were carried out with air containing u p to 0.5% v/v Son; throughout a run gas samples were taken from the reactor inlet and outlet streams and analyzed for SO? with an infrared gas analyzer (Grubb-Parsons Model 30). A run was continued until t h e outlet concentration of SOL became greater than about 90% of its inlet value; this occurred generally within 2 h of start-up. T h e airSO?stream was then turned off, the reactor was purged with nitrogen, and then methane was passed through to reduce the copper sulfate back to oxide and to release the SO2 which was led to waste. Series of runs were performed with four different packed bed depths (4.8,12,and 16 cm), and a t flow rates ranging from below minimum fluidization velocity, Vm*.u p to about 3c',f; bed temperatures were varied over the range 300-430 "C. A number of subsidiary experiments were carried out to measure the degree of expansion of the fluidized bed a t various gas flow rates; these data were needed in order to determine the extent of bubble formation within the bed. Theoretical Model. From t h e large number of fluidized bed reactor models available in t h e literature (Yates, 1975), that developed by Partridge and Rowe (1966) was chosen as being suitable for obtaining predictions to compare with the results of this study. The model is based on the theory that the gas flow into a fluidized bed is divided into two phases, the interstitial or dense phase and the bubble phase, and that the bubble phase accounts for all flow in excess of t h a t required to just fluidize t h e bed (Davidson and Harrison, 1963). Back-mixing in t h e dense phase is considered t o be negligible. Assuming the kinetics to be of the first order with respect to the reactant gas, A, a mass balance on A for a n element of t h e interstitial phase of height, d h , gives
where U I is the interstitial gas velocity, C.AJand CACare the concentrations of reactant gas in t h e two phases, QF, is a n inInd. Eng. Chern., Process Des. Dev., Vol. 16, No. 3, 1977
347
Table I. Solid Absorbent Properties (a) Chemical Analysis (weight %) 94.31 CUO 5.22 C 0.38 Si02 0.008 Fez0 0.005 Na2O 0.004 Ti02 0.085 (b) Physical Properties '41203
""u A14
N,
Figure 1. Fluidized bed flow svstern (schematic).
Pure alumina
terphase gas flow rate (define(. by eq 13), A I and Ac are the areas of the bed cross section occupied by the two phases, k I is the interstitial phase reaction rate constant, and h is the height coordinate. The parameters L'1, QE, Ac, AI, and hI are dependent on the fluid dynamics of the bed. T h e corresponding equation for the bubble phase is
These two equations can be solved simultaneously by combining them to form a single linear second-order differential equation d2C,\ dh
+
dC4 dh
- J-
+ KCA = 0
Mean particle size, mm Bulk density, g/mL Voidage Pore volume (total) mL/g Pore volume (pores
