You are on page 1of 8

6

Ind. Eng. Chem. Fundam. 1981,

20. 6-13

Study of Kinetics of Carbon Gasification Reactions


Satyanarayana Katta' and Dale L. Kealrns
Chemical Engineering Research, Westinghouse Research and Development Center, Piltsburgh, Pennsylvania 75235

The rates of carbon-steam (C-H20) and carbon-carbon dioxide (C-COp) reactions were studied in a laboratory f 920 to 1040 O C at 1013 kPa. Rate equations suggested fluidized bed with coke breeze over a temperature range o by a simplification of the Langmuir-Hinshelwood rate expression and also Ergun's model for a fully back-mixed fluidized bed were developed for both reactions, with rate as a function of temperature and gas composition. The success of these equations in fming gasification results affords some support for Ergun's model but not in all details, and no attempt is made to prove the detailed mechanisms for the reactions in the present study. The reaction h e inert gas concentration. The variation of the reaction rate was found to be independent of the particle size and t rate with carbon conversion was also studied. This study forms a basis for interpreting char reactivities to be used

in projecting plant performance within the range of experimental conditions studied here.

Introduction The objective of the present study was to develop rate equations of practical value for carbon-steam (C-H20) and carbon-carbon dioxide (C-C02) reactions and to use them for predicting the gasification rates in pilot- and commercial-size gasifiers. This study is not intended to obtain a fundamental equation for carbon gasification or to verify a theory. Rather, it is to use a theoretical basis for correlating the data. Further study is needed to obtain a fundamental equation. This study forms part of a continuing investigation of coal behavior in connection with the development of the Westinghouse coal gasification process. Future studies will consist of two parts: (1)the initial reactivity of several chars and the variation of reactivity with carbon conversion; (2) a gasification model in which the rate equations and char reactivity will be used for designing the gasifier and for identifying optimum operating conditions. A review of the recent studies on carbon gasification reactions shows that some of the equations presented in the literature contain a large number of parameters. The investigators generally expressed severe criticism about the use of a large number of parameters, since any form of rate equation can be fitted to the experimental data. Such an equation may not have any relevance to the reaction mechanism. Rate equations based on limited data and assumptions made on the nature of the reactions must be examined carefully. Following is a brief review of some of the models given in order to show the conflicting results of these reactions and the need to conduct additional studies. Von Fredersdorff (1955) noted that for the C-C02 reaction a number of consecutive reversible and irreversible steps have been employed in various combinations in the literature to derive rate equations. He derived four rate equations based on various reaction mechanisms and evaluated each of them by employing his experimental data. He presented the following rate equation, generally known as a Langmuir-Hinshelwood rate expression, on the basis of his experimental data (1) Y = kPPcoJ(1 + W C O + PPCOJ where k is a rate constant and a and 0 are equilibrium constants. He showed that the same form of rate equation can be derived from different mechanisms for the carbon gasification reactions, showing that the reaction mechanism is not uniquely defined by a specific rate equation. Simple rate measurements are inadequate for evaluating a theory.

May et al. (1958) studied coke gasification in fluidized and fixed beds and presented correlations for design purposes. They derived rate constants from the results on the fixed-bed reactor with the assumption of plug flow. Using these rate constants, they found the steam conversion in the fluidized bed to be very close to that predicted for a completely mixed bed, despite the lengthto-diameter ratio of about 20. The flow in this fluidized bed can be expected to be close to plug flow rather than back-mix flow. They attributed this agreement to the presence of bubbles in a fluidized bed. They showed, also, that the water-gas shift reaction failed to reach equilibrium at low steam conversion and exceeded equilibrium at high conversion. Squires (1961) analyzed the data of May and others. He concluded that in fixed beds carbon activity is highest a t the inlet because of high steam partial pressure and low hydrogen partial pressure (lowest at the outlet), but that in fluidized beds the carbon activity remains relatively constant because of periodic reactivation of the entire carbon surface. He criticized the use of a fluidized bed for obtaining differential rate data for various steam/hydrogen ratios because of the effect of the latter on carbon activity. The magnitude of this effect is unknown. Ergun (1962) conducted a comprehensive study on carbon gasification reactions. He postulated a reaction mechanism for both C-H20 and C-C02 reactions and derived rate equations that agree well with his experimental data. In spite of this agreement and the critical evaluation of the effect of various factors he made on the experimental data, his model has largely been ignored in the literature. He studied the kinetics of graphite and metallurgical coke only. Ergun (1955) had mentioned that the concept of the order of reaction in carbon gasification study was controversial and that apparent orders were largely empirical factors. Turkdogan and Vinters (1970) postulated a reaction mechanism involving two rate-controlling reactions in series-namely, the dissociation of COPand the formation of CO on the surface of carbon, for which the reaction rate is proportional to the partial pressures of C02 and to the square root of the partial pressure of COz, respectively. They found that at low CO content the rate is proportional to the square root of the partial pressure of C02 and, in the presence of more than 10% CO, that the rate is proportional to the partial pressure of COz. Rao and Jalan (1972) presented a critical evaluation of several theories and rate equations for the C-C02reaction. They concluded that a two-stage mechanism involving

0196-4313/81/1020-0006$01.00/0 0 1961 American Chemical Society

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981 7

oxygen exchange between the carbon surfaze and the gas phase, followed by the rate-limiting carbon gasification step and a LangmubHinshelwood rate expression, represented a large part of the published data. They obtained an activation energy of 79.6 kcal/g mol for the C-C02 reaction. Fuchs and Yavorsky (1975) investigated the reaction rates of hydrane and synthane chars with H 2 0 and C02. They analyzed the results according to Ergun's model, and their conclusions are essentially in agreement with his. They obtained the same gasification rate constant for both reactions and reported an activation energy of 56 kcal/mol for both reactions. Dutta et al. (1977) studied the reactivities of Illinois No. 6 coal, Pittsburgh HVab coal, and chars made from both in a C 0 2atmosphere. They correlated the change of reaction rate with carbon conversion, using a parameter that is the ratio of available pore surface area at any stage of conversion to the initial available pore surface area. They correlated the char reactivities with the oxygen contents of parent coals with reasonable success and found the reactivities to be almost proportional to the surface areas occupied by pores above approximately 15 A in radius. They found the activation energies of all the materials they studied to be the same at a value of 59 kcal/mol. Johnson (1974) presented correlations for char gasification kinetics from extensive experimental data obtained with a thermobalance and a fluidized bed for a temperature range of 1089 to 1366 K and a pressure range of 101.3 to 7090 kPa. His idealized gasification model consists of three consecutively occurring stages of devolatilization, rapid-rate methane formation, and low-rate gasification. The last was assumed to consist of three reactions: carbon-steam, hydrogasification, and a third reaction, the stoichiometric sum of the first two. The rate expression contains the rate constants for each of the three reactions, a surface area term, and a surface reactivity term, both of which depend on the carbon conversion. These equations have been widely used for modeling by various investigators. The main drawbacks of Johnson's model are the complexity of its rate equations and its use of a large number of parameters. Its success in predicting the performance of larger units has yet to be demonstrated. The correlations Johnson presented are applicable only to gases containing steam and hydrogen. Jensen (1975) studied the C-H20 reaction in a fluidized bed in the temperature range of 1040 to 1430 "C at atmospheric pressure. He concluded that the shrinking-core model is applicable for the kinetics of the reaction. He did not study the effect of gas composition on the reaction rate and seems to have ignored it in obtaining the rate equation as a function of temperature. He reported an activation energy of 19.8 kcal/mol. Linares et al. (1977) studied the reactivities of several chars in air, H20, and C02. They correlated their results with carbon conversion as a function of t / 7 0 . 5 , where t is the reaction time and 70.5 corresponds to the time for 50% carbon conversion. They successfully normalized their gasification data at one temperature and gas composition plots of almost similar into carbon conversion versus t / ~ 0 . 5 shape. Different reactions were closely described by a first-order rate equation over a carbon conversion range up to 0.7 and by a cubic equation over the entire range. It would be interesting to see if similar correlations could be obtained over wide ranges of temperature and reactant concentration. Although the initial rate of carbon gasification reactions has been investigated in detail, the change in the rate of

gasification with carbon conversion has been studied only to a very limited extent and has not been very well established. A careful review of the literature shows that there still exists a need for additional data on carbon gasification reactions for different types of chars and, also, for analyzing the results in terms of a meaningful model established on the basis of a comprehensive study. Nature of Carbon Gasification Reactions The following are the important reactions in the gasification of carbon C + H2O CO + H2 (2) c + c02 2 c o (3) CO + H2O

-F?

COP + H2

(4)

On the basis of information available in the literature, the rate of reaction of char with H 2 0 and C 0 2 appears to depend on the following factors: (a) the nature of the carbon (i.e., the type of coal from which char is formed) and the temperature used in the preparation of the char; (b) the change in reactivity during gasaification (increase or decrease in the effective surface area available for reaction); (c) the temperature and pressure of the reactions; (d) the gas composition (concentration of steam, C02, CO, and H2); (e) the mineral content of the char; and (f) the particle size if the reactions are diffusion controlled. It is apparent from a large number of studies that the carbon gasification reactions are chemically controlled below a temperature of about 1100 "C for the particle sizes normally used for gasification (Von Fredersdorff, 1955). Above this temperature level the diffusion effects become important. Another important feature of the reactions is that in the chemical-reaction-controlled regime, carbon conversion takes place throughout the particle, s cifically at char surfaces located within micropores ( 6 5 ) and not just at the surface or reaction interface (Johnson, 1975). Dutta et al. (1977) stated that for the C-C02reaction the surface area occupied by pores smaller than 30 A in diameter is unavailable for reaction. From a detailed study of the pore structures of various coal and char samples, Dutta et al. (1977) outlined the structural changes that take place as the reaction proceeds. They observed that the dimensions of the solid particles remain practically constant up to a conversion of about 80%. At higher conversions the solid disintegrates into smaller fractions. The initial pore size distribution and its variation with carbon conversion would be the most important factors in controlling the reactivity of any carbonaceous material. Experimental Work The carbon gasification reactions were conducted in a reactor of 3.5 cm i.d. and 30.5 cm height heated externally by an electric furnace, a schematic diagram of which is shown in Figure 1. A sample of about 35 g of coke breeze -1.0 +0.25 mm in size (18 X 60 mesh) was placed on the distributor and fluidized by nitrogen. Bed weights of 10 and 20 g were used in a few of the tests to determine the effect of bed weight on reaction rate. The reactor was pressurized to 1013 kPa and heated to higher than 1040 OC. When steam was used the first condensate indicated the beginning of the reaction. The bed temperature was stabilized at about 1040 "C, and a gas sample was taken. Then the bed temperature was decreased, and gas samples were taken at temperature intervals of 30 "C. Inlet gas samples were also taken. At the end of the reaction, the bed material was weighed and the product gas line was flushed to collect fines. The amount of fines collected was always very small.

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981

Table I. Analysis of Coke Breeze, Wt % ultimate analysis


C

proximate analysis volatile matter fixed carbon moisture ash 3.0 84.5 0.8 11.7

H 0 N ash Table I1
Bed

85.0 0.7 0.8 0.9 11.7

Mixture

temperature pressure partial pressure partial pressure partial pressure partial pressure gas flowrate

Operating Conditions 920-1040 "C 1013 kPa of of of of Inlet Conditions H, 0.0-302.8 kPa H,O 109.4-342.3 kPa CO 0.0-213.7 kPa CO, 202.6-413.2 kPa 23.7 L/min a t 101.3 kPa and 16 "C

S.G.

- Steam Generator

Condensate Receiver

Figure 1. Flow diagram of laboratory fluidized-bed test unit.

The steam generator consisted of an insulated stainless steel boiler containing an immersion heater, a thermocouple well, and a demister. The rate of steam generation was controlled by regulating the electrical power to the heater, measured with a precision wattmeter. A constant voltage transformer was used to stabilize the house power supply. The steam generator was calibrated in a series of runs by measuring the rate of steam generation at different power levels and by determining the correction for water vapor loss during the heating-up period in separate tests. The steam generation rate was found to be independent of the water level in the boiler. The reactor was made of Inconel 600 alloy. A perforated distributor plate and a coiled tube for preheating the fluidizing gas were part of the reactor. The chromel-alume1 thermocouples contained in a single Inconel protection tube were used to measure and control the bed temperature. One thermocouple was connected to a precision potentiometer, and the other led to an on-off temperature controller. The reactor and a furnace to heat it were contained in a pressure vessel. The furnace consisted of three heaters, each of which was controlled by a variable transformer and monitored continuously with a thermocouple. Constant pressures were maintained in the reactor and the pressure vessel by two back-pressure controllers. Samples of off-gas for analysis were collected in evacuated stainless steel sample cylinders that were connected to a manifold located downstream from the back-pressure controller. The gas samples were taken by opening the bottles momentarily by means of solenoid valves operated with manual switches. The rate for the C-H20 reaction at any temperature was determined from the total amount of CO and COz present in the product gas and the estimated amount of carbon in the bed at that time. The reaction rate for C-C02 was calculated from the difference in the amount of CO present in the product and inlet gas. The reaction rate is defined as the weight of carbon reacted per unit time per unit weight of carbon in the bed. The amount of carbon consumed during any period was estimated from the arithmetic average of the reaction rates at the beginning and the end of the period. This checked very well with graphical integration over the same interval. Any small error that might have been introduced by this method is of no practical consequence, as the initial weight of the sample was about 35 g and the carbon conversion in any test was less than about 10%. Gas analysis was performed on a Basic Gas Chromatograph Model 8000 made by Carle Instruments. Measurements on pore size distribution were made with a

Data on Fluidized Bed Pittsburgh coke breeze (-1 +0.25 mm) bulk density 0.16 g/cm3 static bed height 4.8 cm minimum fluidization velocity 8.4 cm/s a t 1800 "F and 1013 kPa superficial velocity at 17.0-18.8 cm/s operating conditions mode of fluidization bubbling bed material
1

0 it

0 02

0 04 0.M 0.08 Square of Cartnn Conversion

0 . 10

0.12

Figure 2. Reaction rate as a function of carbon conversion for C-H,O reaction.

Micromeritics mercury penetration porosimeter Model 910 series with a pressure range of 0 to 206700 kPa. The surface area was measured by using COz as the adsorbate on a Micromeritics Model 2100 surface area analyzer and employing the Dubinin-Polanyi equation for analyzing the results. The ultimate and proximate analyses of Pittsburgh seam coke breeze used in this study, as percent by weight, are given in Table I. Coke breeze was chosen as a reference material because it had been used for the initial tests on the process development unit, is less reactive than chars, and, thus, is easier to use for a kinetic study. We pretreated this material to remove tar by heating it up to a temperature of 982 "C and then allowing it to cool to room temperature in a nitrogen atmosphere. The surface area and pore volume of coke breeze covering the pore size range of 0.01 to 100 pm diameter measured about 13.9 m2/g and 0.68 cm3/g, respectively. (For operating conditions and fluidized bed data, see Table 11). The effect of carbon conversion on the reaction rate of coke breeze with steam at temperatures of 927 and 982 "C

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981 9

at an inlet steam partial pressure of 309 kPa in nitrogen is shown in Figure 2. The average values of PHZ/PH8in the product gas at these two temperatures are also shown. The variation of the reaction rate with carbon conversion is similar at these two temperatures. The curve corresponding to 982 "C was used to correct the results on the C-H20 reaction. No correction was applied to the results on the C-C02 reaction, however, as the carbon conversion was very small. Reaction Models According to Ergun's theory, the first step of the C-H20 and C-COz reactions on the surface of carbon may be represented by c, + COB c, + co (5)

0.61

0.d
0.4

/
0.5

-0.1

1.0

1.5

2.0

2.5

3.0

Inverse Rertion Rate x

I O ' ,

min

(6) where C, and C, represent a free site and an occupied site possessing an oxygen atom, respectively. These reactions are denoted as oxygen exchange reactions. The second step of the reaction involves the transfer of carbon from solid phase to gas phase and is represented by the following for low conversions c, co + c, (7) Ergun postulated that the desorption of CO is the ratecontrolling step in both reactions. He made the following assumptions: the total number of reaction sites (sum of free and occupied sites) is constant; the oxygen-exchange reactions are very fast and attain equilibrium, implying that the water-gas shift equilibrium is maintained in the gas phase; the composition of the gas phase inside a particle is uniform and identical with that surrounding the particle. He derived the following equations for C-C02 and C-H20 reactions, respectively (8) 7 1 = k l / U + PCO/KlPCO~)
C,

+ H2O ~t C, + H2

Figure 3. Determination of reaction rate parameters for C 4 0 2 reaction.


0.7r
0.61
A

0.2
0.1

1
1
O.7r

/
/A

Temp = 10IODC

-0. I

2
Inverse Reaction Rate x

4
min

I O ' ,

Figure 4. Determination of reaction rate parameters for C-C02 reaction.

YZ = k 2 / 0 + PH,/&PH,O)

(9)

where kl and k2 are rate constants that depend on the temperature and type of carbon, Kl and K2 are equilibrium constants of the oxygen exchange reactions, and y1 and y2 are reaction rates per unit mass. These equations are not applicable to experiments under vacuum. For complete mixing in the reactor, the ratios PH,/PH,O and pco/pco for the individual reactions remain constant throughout the reactor, and the reaction rate expressions for an isothermal fluidized bed can be written as (10) Pco/Pco, = -K1 + Klkl(l/YJ A linear relationship between pco/pcoz or PHz/PH 0 and the inverse reaction rate would support the validity of Ergun's rate equation. Ergun also presents rate expressions, different from (10) and ( l l ) , for a fluidized bed with negligible back-mixing of gas. These expressions were not used in the present studies. According to the theory, the rates of carbon with H20 and C02should be the same since the rate-controlling step is the desorption of the carbon-oxygen complex from the carbon surface. Ergun obtained higher gasification rates (- 60 percent) in the case of H20, which he explained in terms of a higher number of reaction sites than for COz. The water-gas shift reaction did not reach equilibrium in some of his experiments. He suggested two explanations: (1)oxygen exchange reactions are not extremely fast as

"'I
0.4

/
Temp= 9 W C 2 4

/
0

1 0 1 2 1 4

Inverse R e r l i a n Rate x I O ' ,

min

Figure 5. Determination of reaction rate parameters for C-C02 reaction.

postulated; (2) some sort of hindering due to pore diffusion which is not manifested by either dependence on particle size or inert gas concentration. Ergun's theory waa not nearly as successful for mixtures of H20-C02, H20-H2, or H20-CO as it was for pure reactants. Results CarbonXarbon Dioxide Reaction. The experimental data of the C 4 O Zreaction were plotted with p c o / p ~ as ~, a function of the inverse reaction rate a t temperatures of 1040, 1010,980, and 950 "C and are shown in Figures 3 to 6, respectively. The agreement between the results and Ergun's rate equation is very good. The intercept on the abscissa gives the value of l/kl, and the negative intercept on the ordinate gives the value of K1. The lines were

10

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981


0.7,

Temp= IW'C

2
x

Inverse Relrlion Rate

0.2

0.4

0.6

0.8

1.0
x

1.2

1.4

1.6

I O ' .

min

lmerse Reation Rate

IO-'

min

Figure 6. Determination of reaction rate parameters for C-C02 reaction.

Figure 9. Determination of reaction rate parameters for C-HzO reaction.


1.2r
0

1 .5

' c

1.0

0 .5
0.4 0.3

0. 1 0.74

\;;;
0.76

-0.21

1.0

2.0
Inverse Rection Rate x min

I. 0

4.0

ais

am

0.82

aM

0.86

Inverse Absolute Temperalure x lo3, K-'

Figure 10. Determination of reaction rate parameters for C-HzO reaction.

Figure 7. Reaction constant as a function of temperature for C C O z reaction.


1.01

o,2t\
0.02
0.74

0. 3

i
am
0.82

0 .8
O N
$ N

0.6

0 .4

/
I

Temp = 9WC

0 .2
0

I
I
I

i
I I

Inverse Reaction Rate x I O ' .

min

0.76

0.78

0.84

Figure 11. Determination of reaction rate parameters for C-H20 reaction.

Inverse Absolute Temperature x IO3, K-'

Figure 8. Equilibrium constant as a function of temperature for C-COP reaction.

obtained from a regression analysis of the data. It should i n e a r plot be kept in mind that the results are shown on a l as opposed to a conventional logarithmic plot for showing the effect of reactant concentration when observing the agreement between the data and the rate equation. The variation of the reaction constant and the equilibrium constant with temperature is shown in Figures 7 and 8, respectively. The following equations were obtained k l = 0.55 X 1Olo exp(-69000/RT) (12) (13) where k l is the reaction constant, min-', and T is the

temperature in Kelvins. Hence, the reaction rate of C-C02 reaction can be represented by Eq 8. Carbon-Steam Reaction. The results obtained for the C-H20 reaction were analyzed in the same way as those of the C-C02reaction. Data at temperatures of 1040,1010, 980,950, and 920 "C are shown in Figures 9 through 13, respectively. The results and the form of the rate equation derived from Ergun's theory agree very well for this reaction, also. The variation of the reaction constant and the equilibrium constant as a function of temperature is shown in Figures 14 and 15, respectively. The following equations were obtained for the two parameters

K1= 5.86 X lo3 exp(-27000/RT)

k2 = 4.85 X lo6 exp(-482OO/RT)


K2 = 2.25
X

(14)
(15)

lo6 exp(-42600/RT)

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981


0

11

0.8
/

0. 10
I

0.08

Inverse Readon Rate x lQ-',min

0.021

0.74

0.76

0.78

0.80

0.82

0.84

1
0.86

Figure 12. Determination of reaction rate parameters for C-H20 reaction.


0.6
I
I

Inverse Abxllule Temperature x IO3, K-'

Figure 15. Equilibrium constant as a function of temperature for C-H,O reaction.

0'51
0.1

/ "1
X

3 t

0. I

/
I I
I

I2

16

20
min

24

28

32

Inverse Reaction Rate x I O ' .

Figure 13. Determination of reaction rate parameters for C-H,O reaction.

0.21

IO

m
Initial Bed Weight. g

30

Figure 16. The effect of bed weight on reaction rate at different temperatures for coke breeze-steam reaction.

Inverse Abrolule Temperature x IO? K-'

Figure 14. Reaction constant as a function of temperature for C-H20 reaction.

where It2 is the reaction constant, min-', and T is the temperature in Kelvins. The rate of C-H20 reaction is therefore given by eq 9. The effect of particle size was investigated by studying the reaction rates of two particle size distributions of coke breeze, namely, -1 +0.85 mm (18 X 20 mesh) and -1 +0.25 mm (18 X 60 mesh) at 982 "C and an inlet steam partial pressure of 303.8 kPa as a function of carbon conversion. There was no significant difference between the two sizes. The effect of bed weight on the reaction rate of coke breeze-steam reaction was studied by using bed weights of 10 and 20 g in addition to the 35 g bed that was normally used for the same inlet gas composition (PHzO = -202.6, pHz = 0.0 Wa). The results are shown in Figure 16. Note

that the bed weight significantly affects the reaction rate, the rate increasing with a decrease in the bed weight, as can be expected. When we plotted these results according to Ergun's model, we could not observe the effect of bed weight. Some of the past investigations (Zielke and Gorin, 1957) had assumed that a true reaction rate could be obtained by extrapolating the reaction rates at different bed weights to zero bed weight, but a comparison of the reaction rates at zero bed weight obtained by extrapolation with the reaction constants obtained from Ergun's rate equation shows that the former are substantially lower than the reaction constants. This, in turn shows that the practice of extrapolating reaction rates to zero bed weight is not desirable. The rate equations presented here are applicable for initial reaction rates only. The variation of reaction rate of the C-H20 reaction with carbon conversion is shown in Figure 2 at temperatures of 982 and 927 "C for the same inlet gas composition. The same variation can be expected to be valid for other gas compositions as well. Johnson (1974) presented correlations for y(1- X)'I3 vs. X 2 with pHB/pHz as a parameter. The lines for the same pressure are nearly parallel, showing that the effect of carbon conversion on the reaction rate is independent of the gas composition. The characteristics of pore surface area, for example, can be expected to influence the effect of carbon conversion on the reaction rate. The initial increase in the reaction rate with carbon conversion can be explained as due to the opening up of pores which, in turn, results in an increase in the internal surface area. The subsequent

12

Ind. Eng. Chem. Fundam., Vol. 20, No. 1, 1981

decline in the reaction rate can be attributed to a decrease in the internal surface area. Discussion The low values of the ratio of superficial velocity to the minimum fluidization velocity and the bed-height-to-diameter ratio of about 1.4 would lead one to expect the fluidized bed used in the present study to be in a bubbling, not in a slugging, region. By estimating the bubble rise velocity, we did indeed find the bed to be in the bubbling region. Hence the assumption of back-mix flow appears to be reasonable. A comparison of the results of both reactions shows that the rate constants of the C-H20 reaction are about 2.5 to 5 times higher than those of the C-C02 reaction over the temperature range of 950 to 1040 C. Ergun (1962) obtained different rate constants for these two reactions, although according to his theory the same rate constants are expected. He explained this difference by reasoning that H 2 0 is capable of reacting with a greater number of reaction sites than is COP A comparison of the results of this study with those of Erguns study for either reaction is difficult. The ash content in the coke breeze used here is 11.770,as opposed to 7.4% in the coke Ergun used. A valid comparison between the two materials cannot be made without information on the ash composition and pore surface area. Also, the experiments in the present study were conducted at a pressure of 1013 kPa, while Erguns tests were performed at atmospheric pressure. Possible alternative explanations for the different activation energies obtained for the two gasification reactions are that the flow in the reactor deviated from back-mix flow and that the same reaction step is not controlling in both cases. If the latter is true, minerals present in the coke breeze used in this study would be expected to catalyze the two reactions to different extents. It is unclear which of these factors contributed to the appreciable difference in the activation energies of the two reactions. In a study on the effect of mineral matter content on the rate of C-O2 reaction, Jenkins et al. (1973) found that the amount of calcium and magnesium present in chars had the greatest effect on the reactivity, while potassium, sodium, and iron had none. The rectivity of a char may not be the same in an oxygen, HzO, or C02 atmosphere since the mineral matter content influences each of the carbon gasification reactions to a different extent (Tomita et al., 1977). The ratio of the equilibrium constants of the C-H20 reaction to that of the C-C02 reaction is not equal to the equilibrium constant of the water-gas shift reaction in the present study. We believe that the water-gas shift reaction failed to attain equilibrium at the exit of the reactor, as the steam conversion was very low. This conclusion is supported by the work of May et al., Ergun, and Squires. In the present study, steam decomposition varied from 0.1 to 20%, while in Erguns study, it varied from 11to 99.6%. The observation that the water-gas shift reaction does not reach equilibrium at low steam conversion is certainly a drawback in Erguns theory. The present correlation may be regarded as supporting the form of the Erguns rate equation rather than its theory. Erguns results show that at steam conversion of less than 20%, the water-gas shift reaction is far from equilibrium and suggest that the oxygen exchange reactions are not very fast. In most of Erguns tests, steam conversion was fairly high (more than 50%),and the measured equilibrium constant of the oxygen exchange reaction of the C-H20 system agreed well with that derived from the C--C02system. Such would not be the case if data from

a few of the tests, where steam conversion was low, were used in the data analysis. This observation leads to the possibility that the equilibrium constant depends on the extent of steam conversion. In commercial gasifiers one can expect the gas flow to be closer to plug flow than to back-mix flow and the steam conversion to vary from 0 to more than 60%. This finding implies that the water-gas shift reaction reaches equilibrium only near the top of the bed. The rate equation to be used for the design or prediction of performance of a reactor should, therefore, contain parameters that are independent of steam conversion. For design purposes, however, the reaction rates are insensitive to the values of the equilibrium constant. The advantages of Erguns theory are the simplicity of the rate equations and the fundamentals from which they were derived. The rate equations contain only two parameters: a rate constant and an equilibrium constant. The activation energies of the C 4 0 2 reaction for various carbonaceous materials appear to be the same (Fuchs and Yavorsky, 1975), and, hence, the rate constant for any material can be established by measuring the reaction rate at one temperature. The variation of the equilibrium constant, formulated by Erguns theory, with temperature has not been established for different materials. We have determined and reported elsewhere (Katta and Keairns, 1980) the reaction rates of various chars (Minnehaha, Renton, Utah, and Western Kentucky) with steam. These results also show a linear relationship bethus showing tween the inverse reaction rate and pH2/p~&, the validity of Erguns rate equation for different types of carbon. The reactivity of a carbonaceous material for a coal gasification process can thus be estimated by studying its reactivity in steam, since most of the carbon conversion takes place through the char-steam reaction. One can infer from the analysis that the inert gas concentration has no influence on the rate of either reaction, and, hence, it is meaningless to express the rate as a function of the partial pressure of the gaseous reactant as is commonly done in studies on carbon gasification. This study has shown clearly that the practice of obtaining the reaction rate at different bed weights and extrapolating these results to zero bed weight to get true rate is incorrect. The rate equations we developed in this study are applicable for initial reaction rates only, as are the equations presented in most studies. The variation of the reaction rate with carbon conversion must be determined separately for each type of carbon. The method established here provides a reliable basis for comparing the reactivities of chars by using a simple test. An average reaction rate of char can be obtained by integrating the reaction rate over the range of carbon conversion. The reaction rate of any char at a given temperature and gas composition is obtained from the product of the initial reaction rate of coke breeze as calculated from the rate equations, the initial relative reactivity of the char with reference to coke breeze, and the ratio of the average rate to the initial rate of the char. Hence, in order to design a gasifier or to predict the performance of a gasifier with a given configuration, the rate equations presented here should be used along with the initial reactivity of the feedstock and the information on the variation of its reaction rate with carbon conversion in a gasification model.
Conclusions Rate equations were developed for the C-H20 and CC02reactions with coke breeze employing a fluidized bed. Results agree well with the form of the rate equation suggested by a simplification of Langmuir-Hinshelwood rate expression as suggested in the Appendix and also

Ind. Eng. Chem. Fundam., Vol. 20,

No. 1, 1981 13

Erguns model. The values of the equilibrium constants, however, and the activation energies of the rate parameten do not agree with those reported by Ergun. We believe that the water-gas shift reaction equilibrium is not maintained at the exit of the fluidized bed because of the low steam conversion in this study. We have presented possible reasons for the above discrepancies and the differences between Erguns model and our results. The rate constant of the C-H20 reaction was about 2.5 to 5 times higher than that of the C-C02reaction in the temperature range studied here, and the activation energies were 48.2 and 69 kcal/mol, respectively, for the two reactions. There is a considerable difference in the activation energies of the reaction and the equilibrium constants for both reactions, although the form of the rate equations is the same. The variation of reaction rate with carbon conversion was also studied. The reaction rate was found to be independent of the particle size. The advantages and merits of Erguns model and the drawbacks of some of the other models and studies presented in the literature have also been examined here. This investigation provides a reliable method to determine char reactivities and the information needed for predicting gasifier performance. Acknowledgment This work was performed as part of the Westinghouse coal gasification program under DOE Contract EF-77-C01-1514. The authors appreciated the constructive comments and suggestions by the reviewers of the manuscript. Appendix The following mechanism was presented by Van Fredersdorff (1955) for the C-C02 reaction where the equilibrium adsorption of C02on the carbon surface is followed by surface reaction and the adsorbed and gas phase products are in equilibrium

This leads to the rate expression


= kfgH,O/(l
+ apH, + PPHzO)

(A91

where kf = klkz/kl, LY = k3//k3, and P = kl/kl. This equation can be reduced with the same reasoning as above to the following

Y = k 2 0 c / ( l + WH,/PPH,O)

(A101

(CO,)

+ c -2 co + (CO) ks (CO) co

(A2) (A3)

which is also of the same form as that presented by Ergun. Examination of the model on which the LangmuirHinshelwood equation was based reveals that the watergas shift reaction equilibrium was not assumed (Van Fredersdorff, 1955). The form of the Ergun rate equation could be derived from the Langmuir-Hinshelwood equation by assuming that the adsorption of the gaseous reactant on the carbon surface and the desorption of the products are very fast, as shown earlier. A comparison of the term ( d p m + ppcq) with unity cannot be made either for Erguns data or for the present data, as the values of a and p are not available. Ergun, however, reported the ratio of p to a,which is the equilibrium constant. Nomenclature kl = reaction rate constant for the C-C02 reaction, min- kz = reaction rate constant for the C-HzO reaction, min- K1= equilibrium constant of the oxygen-exchange reaction for the C-C02 reaction K2 = equilibrium constant of the oxygen-exchange reaction for the C-H20 reaction PCO, P C O ~ P , H ~ ,PH,O = partial pressures of CO, COz, H2, and HzO, respectively y1 = reaction rate per unit mass of the C-C02reection, m i d yz = reaction rate per unit mass of the C-H20 reaction, min-l T = temperature of the bed, K X = fractional carbon conversion a, = active surface area available for reaction per unit weight of carbon Literature Cited
Dutta, S.; Wen, C. Y.; BeR, R. J. Ind. Eng. Chem. ProcessDes. Dev. 1977,
16, 20-30. Ergun, S. Ind. Eng. Chem. 1955, 47(10), 2075-80. Ergun, S. Bur. Mlnes Bull. 1962, No. 598. Fuchs. W.; Yavorsky. P. M. presented at Symposium on Structure and Reactivity of Coal and Char, 170th National Meeting of the American Chemical Society, Chicago, Aug 1975. Jenkins, R. G.; Nandl, S. P.; Walker, P. L., Jr. Fuel, 1973, 52, 288. Jensen, G. A. Ind. Eng. Chem. Process Des. Dev. 1975, 14, 308-314. Johnson, J. L. Adv. Chem. Ser. 1974, No. 131, 145-178. Johnson, J. L., Presented at Symposium on Structure and Reactivity of Coal and Char, 170th National Meeting of the American Chemical Society, Chicago, Aug 1975. Katta, S.; Keairns, D. L. Char Reactivities and Their Relationship to Pore Characterlstics, presented at 180th National Meeting of the American Chemical Society, Las Vegas, Aug 1980. Llnares, A.; Mahajan, 0. P., Walker, P. L., Jr. presented at the 173rd National Meetina of the American Chemical Society, ~. Division of Fuel Chemistw. .. New Oyleans, 1977. May, W. G.; Mueiier, R. H.; Sweetser, S. B. Ind. Eng. Chem. 1958, 50, 1289. Rao, Y. K.; Jalan, B. P. Metall. Trans. 1972, 3 , 2465-2477. Squires, A. M. Trans. Inst. Chem. Eng. 1961. 39, 10. Tomita, A.; Mahajan, 0. P.; Walker, P. L., Jr. Presented at the 173rd National Meeting of the American Chemical Society, Division of Fuel Chemistry, New Orleans, 1977. Turkdogan, E. T.; Vinters, J. V. Carbon 1970, 8 , 39-53. Von Fredersdorff, C. G. Inst. Gas Technol. Chicago Res. Bull. 1955. No. 19. Zielke, C. S.; Gorin, E. Ind. Eng. Chem. 1957, 49, 396.

(the parentheses indicate adsorbed species). This mechanism leads to the rate expression
Y = k~cPco,/(l+ .Pco + pPC0,)

(A4)

where k; = klk2/kl, a = ki/k3, p = kl/k(, and uc is the active surface area available for reaction per unit weight of carbon. If one considers the values of aand p presented in the literature and for the ranges of pco and pco, used in this study, unity can be neglected in comparison with (cupco + ppco,). Dividing the numerator and the denominator by ppco,, one obtains the following equation Y = k2ac/(l + .Pco/pPco,) 645) which is of the same form as Erguns. Von Fredersdorff proposed a similar mechanism for the C-H20 reaction

Received for review October 23, 1978 Resubmitted April 28, 1980 Accepted September 19, 1980