Modeling of Noncatalytic Gas-Solid Reactions

Modeling of Noncatalytic Gas-Solid
Reactions
This paper presents a critical review of the developments in the mathematical P. A. RAMACHANDRAN and
modeling of gas-solid noncatalytic reactions with particular emphasis on recent
trends in the subject. A number of models proposed for analyzing this class of re- L. K. DORAISWAMY
actions have been reviewed with a fairly detailed discussion of the methods of Naibnal Chmmkd Laboratary
incorporating structural changes which occur in the solid with the progress of re- PoocyIndl.
action. The present status on the modeling of various types of complex gas-solid
reactions is reviewed. Also the paper points out a number of areas in which future
research may be needed. The review concludes with a critical discussion on the
type of experimental data necessary for model verification and some comments
on the choice of model for a given system.
SCOPE
Noncatalytic gassolid reactions are encountered in a variety solid properties with the course of reaction. Some of the major
of chemical process industries. Tbe major applications are found developments in this area have been reviewed in a book by
in the fields of extractive metallurgy, control of gaseous pollu- Szekely et .I. (1976) and in an edited monograph by Sohn and
tants, coal gasification processes, combustion of solid fuels, Wadsworth (1979) Since then considerable advances have taken
catalyst manufacture, etc. Tbe mathematical modeling of these place, and it is felt that there is a need for a comprehensive and
systems is important in order to interpret laboratory data critical review of the major recent developments in this area,
on these systems and in design and scaleup. The problem is and this review is written with this objective. Such a review also
complex since in addition to the interplay of heat and mass helps in evaluating some of the trends in research in this
transfer, other considerations are necessary to account for the field.
transient nature of the problem and the effects of changes in
CONCLUSIONS AND SIGNIFICANCE

In this review the various models which are commonly used Another important area of research is in the field of complex
to describe gas-solid noncatalytic reactions have been discussed. gas-solid noncatalytic reactions. A classification of the impor-
The three common models which have been in use are the sharp tant reactions belonging to this class has been presented and the
interface model, the volume reaction model, and the particle- major developments in tbe modeling of these have been pointed
pellet model. These models have been recently modified to take out. The review also covers special characteristics of gasification
into account the effects of structural changes due to chemical and decomposition reactions and reactions showing significant
reaction and sintering. AIso new models which take into account nucleation effects. The information on the stability of nonca-
the basic porous nature of the solid and the pore size distribution talytic gas-solid reactions is also breifly reviewed. Finally the
have been proposed. The review covers all these major devel- review comments critically on the type of experimental data
opments. Further, in each case the model parameters and their required for model evaluation and attempts a critical qualitative
physical significance have been clearly indicated. comparison of various models.
A general type of gas-solid noncatalytic reaction may be r e p and such reactions are known as decomposition reactions. Finally
resented as another important class of reactions is where only gaseous products
are formed ( uc = 0)and such systems belong to the gasification
u A A(g) + V B B ( s )=& V P p k ) -k U C G(s) (i)
reactions. Special features of modeling of decomposition and
in which a gas and a solid react to give gas and solid products. In gasification reactions are also brought out in this review.
some cases a gas and solid may yield only solid products ( u p = 0). For modeling the general reactions represented by Eq. i, three
Modeling of these two types of reactions is very similar and dis- types of models have been proposed: (i) in which the solid is as-
cussed in considerable detail in this paper. In some cases a solid sumed to be nonporois, the reaction occumng in a topochemical
reactant decomposes to yield gaseous and solid products (uA = 0 ) manner; (ii) where the solid reactant is assumed to be porous; and
(iii) where the solid pellet is assumed to be composed of nonporous
subunits known as particles or grains. These models are fint dis-
a)ol.1541.81-82WQ T b e A ~ l r d t u l e d C W E n g l m r slWX2 cussed in the following sections.
AK=hE Journal (Vol. 28, No. 6) November, 1982 Page 881

SHARP INTERFACE MOML incorporates the variation of physical properties with temperature
has been developed by Prasannan et al. (1982).
The sharp interface model is one of the earliest models used and The prediction of temperature profiles in the pellet may be the
is well described in standard textbooks on chemical reaction en- key problem in some situations. For example, in catalyst manu-
gineering. Hence we shall confine our attention here to the sig- facture by reduction of metal saltsor in the oxidation of coke from
nificant refinements which have been published recently. The spent catalyst the temperature rise in the pellet has to be restricted
model is mainly applicable to highly nonporous solid reactants and below a permissible maximum in order to prevent catalyst deac-
assumes that the reaction occurs at a sharp interface that divides tivation. Shettigar and Hughes (1972)have developed models for
the exhausted outer shell (ash layer) and the unreacted core of the the problem of coke regeneration. A related problem is encoun-
solid. The model development for an isothermal first order reaction tered in ash agglomeration for reaction of carbon with oxygen and
with equimolal counterdiffusionof reacting gases and product gases steam (Rehmat and Saxena, 1980).The ash gets agglomerated only
is straightforward and leads to an analytical relation for conver- if sufficiently high temperatures are reached within the particle.
sion-time behavior (for example, Wen, 1968;Sampath and Hughes, The theory developed by Rehmat and Saxena can be used to pre-
1973; Szekely et al., 1976). The complexities that need to be ac- dict these conditions.
counted for are due to the effect of bulk flow (nonequimolal The structural changes due to chemical reaction generally lead
counterdiffusion), existence of pressure gradients in the pellet, to a change in the pellet size, and models to account for this have
nonisothermal effects and effects due to structural changes. been developed by Shen and Smith (1963)for isothermal systems
Detailed formulation accounting for the effects of bulk flow has and by Rehmat and Saxena (1977) and Rehmat et al. (1978) for
been presented by a number of authors (for example, Beveridge nonisothermal systems. A model to account for changes in the ef-
and Goldie, 1968; Gower, 1971; Sohn and Sohn, 1980). The last fective diffusivity of gas in the product layer (Dee) with time due
authors present analytical solutions for a particular case when to the process of sintering has been discussed by Evans et al.
diffusion in the product layer is rate-controlling. The influence of (1973).
pressure gradients has been examined by Deb Roy and Abraham The concept of effectivenessfactor commonly used for catalytic
(1974),and Turkdogan et al. (1973). These gradients are sigrufimt systems can be used to account for the diffusional effects in non-
when Knudsen diffusion predominates in the pores of the product catalytic systems. Such an approach was first suggested by Ishida
layer. Deb Roy and Abraham (1974)have given detailed numerical and Wen (1968). It may be noted here that the effectivenessfactor
solutions to the pressure gradients which arise because of the for a noncatalytic reaction is a function of time unlike that for
Knudsen flow existing in the ash layer with small pores and due catalytic systems.
to the reactant gas having a diffusivity different from that of the The sharp interface model has the advantage of mathematical
product. simplicity although it should be used only for relatively nonporous
Many gas-solid reactions do not obey first-order kinetics and solids. The major model parameters are the surface reaction rate
follow fractional order with respect to the gaseous reactant A. constant k, and the effective diffusivity of gas in the ash layer D,c.
Examples are found in the reduction of Cr203 by Hz (Chu and The parameter D,c should preferably be measured from inde-
Rahmel, 1979),oxidation of ZnS (Cannon and Denbigh, 1957).As pendent experiments and compared with the value obtained from
strong adsorption of A on the surface of the solid reactant B is r-t data. This has been illustrated by Gokarn and Doraiswamy
often responsible for this behavior, such systems may equally well (1971, 1972) for the oxidation of zinc sulfide pellets.
be represented by a Langmuir-Hinshelwoodkinetics. For nonlinear
systems, the conversion-time relations have to be obtained by nu-
merical integration and the methods for doing this have been de-
VOLUME REACnON MODELS
scribed by Sohn and Szekely (1972a) for power-law kinetics and
by Ramachandran (1982a) for a Langmuir-Hinshelwood ki-
netics. Model Formulation
An interesting case of zero-order reactions has been reported by
When the solid is porous, the gas can penetrate into the solid and
Simonsson (1979)for the reaction of NH4F with CaC03 particles.
the reaction may now be assumed to take place all over the volume
The modeling of a zero-order reaction has also been presented by
of the pellet rather than at a sharp interface. The models which
this author. For this case, the transition between kinetic and dif-
represent this situation are known as volume reaction models. The,
fusion control is very sharp and the process is either entirely con-
rate of reaction at the interior points would be of course generally
trolled by kinetics or by mass transfer. A combination of first- and
lower than that at the surface due to diffusional gradients. The rate
zero-order kinetics has been proposed for the oxidation of coal by
of reaction per unit volume of the pellet at a radial position r is
Kam et al. (1976). As the reaction was assumed to follow two dis-
represented in this model by the following equation:
tinct pathways, the rate of reaction consisted of two terms: one with
first-order dependence on oxygen concentration and the other rA = k,, A m B" (1)
independent of oxygen. Some aspects of modeling of such systems
have been discussed by these authors. where m and n are the order of reaction with respect to the gaseous
When the reaction is accompanied by large heat effects, non- species A and the solid reactant B respectively.
isothermal models are required. An added complexity arises here Based on this model the equations for mass and heat transfer
due to the fact that the pseudosteady-state assumption which is within a spherical pellet can be represented as:
generally used in the mass balance is not a good approximation for
heat transfer (Beveridgeand Goldie, 1968). Hence, a transient heat
balance equation which includes the heat capacity of the pellet is
needed. A simple model here is to assume that the entire solid is (3)
at a uniform temperature at a given time and the entire resistance
to heat transfer lies in the gas film. This is a reasonable approxi-
mation in many cases and is generally valid when h << k,/R. The
validity of this assumption for gas-solid catalytic reaction has been
db -
_
dT
[
- -Q exp y (1 - i)] (4)
demonstrated by Carberry (1961). For the more general case where Q is the local rate of reaction in terms of dimensionless
equations for predicting the complete transient temperature profile concentrations and is defined as:
in the pellet have been derived by Luss and Amundson (1969). Q=ambn (5)
Solutions of this in conjunction with the equations for the rate of
movement of the reaction interface provide a useful approach for The other dimensionless parameters 4,p, y and dimensionless time
obtaining both r-t and T ( r ) - t patterns of behavior of the pellet. T are defined in Table 1. (The stoichiometriccoefficient V A is taken
A fairly general computer programme for this purpose which also as unity here and in subsequent portions of the text without loss of
AlChE Journal (Vol. 28, No. 6)

Page 882 November, 1982
need to consider the presence of a completely reacted product layer
because mathematically for a first order reaction (n = I) the con-
centration of B at the surface can become zero only at 7 = -.
Parameter Notation Definition The case of m = 0, n = 1 has also been analyzed in detail by
Dudukovic and Lamba (1978b). The model is likely to be appli-
cable when the species A is strongly adsorbed on the solid so as to
make the reaction zero order in A. However it may be worth noting
here that such zero order behaviour is only an approximate r e p
resentation of the system and the true kinetics may be a Langmuir
Hinshelwood model with a large value of the equilibrium constant.
Analytical solutions for Z-7 behaviour for (0,l) case have been
derived by Dudukovic and Lamba (1978b). A related problem of
considerable theoretical interest occurs when the reaction is zero
order with respect to both gas and solid (m = 0, n = 0).This has
been analyzed in a recent paper by Ramachandran and Dorais-
wamy (1982). Here as both A and B can be completely consumed
generality.) For isothermal systems, only the mass balance equa- by reaction the pellet shows a "jumping reaction zone'' be-
tions (Eqs. 2 and 4) need to be considered. In the formulations of havior.
the heat balance equation (Eq. 3) the transient terms have been
included as these terms are often important in determining the
temperature profile and conversion in the pellet. The latter
quantity is defined as:
Z(T)= 3 1' y2b dy (6)

Model Solutlon Technlques
It is appropriate here to discuss some of the recent developments
where y is the dimensionless distance r / R in the pellet. in the solution of the model equations to obtain the conversion-time
The boundary and initial conditions are standard and are given behavior of the pellet. For isothermal reactions which are first
as: order with respect to the gas a special technique has been developed
by Del Borghi et al. (1976) and Dudukovic and Lamba (1978a)
da based on the concept of cumulative concentration. This variable
At y = 1, - = BiM(1 - a )
dY is defined as:
d0
'- = BiH(1 - 0) C,,, = 1 ' a d T (11)
dY
A t y = O , da- = - -d0 - 0 Using this transformation Eq. 2 reduces to the following form for
dY dY isothermal systems with m = 1
and
v2c, = r#PF(C,) (12)
at7=0, 0=1, b=l (10) where
It is seen from the above formulation that for the simple case of F(C,) = 1 - [l - (1 - n)C,]l/(l-") for n <1 (13)
isothermal systems the volume reaction model has four parameters:
k,, D, (effective diffusivity of gas in porous solid B), m and n. =1 - exp (-C,) for n = 1 (14)
Equation 12 is easier to solve for C, than the original coupled
Dlscueslon of Some Speclflc Cares equations. The required boundary conditions are: (1) the condition
of symmetry for C, at the center, and (2) C, = time elapsed ( 7 )
The case of (1,O)order reaction (m = 1, n = 0) has been analyzed
by Ausman and Watson (1962) and Ishida and Wen (1968). These at the surface for the case of no significant external transport re-
sistance. The conversion in the solid is related to C, by the fol-
authors divided the total reaction time into two periods: (a) constant
lowing equation:
rate period, and (b) falling rate period. The necessity for consid-
ering two such separate periods arises because the concentration
of B can actually become zero at the surface after the elapse of a
particular time, whenever the reaction order, n , is less than unity.
(This time is equal to BA-"/[(l- n)k,A,"]when BiM -+ m for an where SF = shape factor of the pellet.
isothermal system.) After this time, a product layer forms near the It may be noted here that Eq. 12 has the same form as that for
surface through which gas A has to diffuse without reacting before a catalytic reaction with a nonlinear rate law. Hence various
it can reach the reactant B . The pellet then comprises two zones: techniques available for the solution of such problems can be used
(a) product layer X < y < 1, and (b) a reaction zone 0 < y < X here and approximate analytical solutions can be derived for a
(Figure 1).The model Eqs. 2 , 3 and 4 are valid only for the reaction number of cases using, for example, a single collocation point a p
zone. In the region X < y < 1 the quantities 52 and b are zero and proximation (Ramachandran and Kulkarni, 1980)or the asymptotic
model has to be solved as a set of coupled differential equations. property of Eq. 12 (Ramachandran, 1982b). These approximate
Also the problem becomes one of the moving boundary value type solutions are fairly close to the numerical solutions and hence are
as the position X moves inward into the pellet with time. When the useful in estimating the model parameters since they avoid re-
rate of reaction is fairly rapid (4 > 5), the concentration of A drops peated numerical solution of the model for each set of assumed
very sharply in the reaction zone which may now be further di- parameter values.
vided into two zones: (b-1) reaction zone, and (b-2) core of com- Approximate analytical solutions can also be derived by methods
pletely unreacted R. This then leads to the three zone model pro- suggested by Sohn and Szekely (1972b) and Sohn (1978) or by using
posed by Howen and Cheng (1969), Tudose (1970) and Mantri et average solid concentration in the pellet (Wen and Wu, 1976).
al. (1976). These models are also referred to as finite reaction zone Other recent studies on model solution techniques are as follows:
mbdels. Mathematically they are very similar to the two zone use of orthogonal collocation on the original set of equations (Bo-
model proposed by Ishida and Wen (1968). The case of (1,l) and ersma et al., 1980), collocation on the transient problem, that is,
(1,0.5)order reaction (m = 1, n = 1 and 0.5)has been analyzed by without using pseudosteady state hypothesis (King and Jones, 1979)
Dudukovic and Larnba (1978a). For (1.1) order case there is no and use of invariant imbedding methods (Bellman et al., 1977).
AlChE Journal (Vol. 28, No. 6) November, 1982 Page 883

I I\
t=o- '
I
ws FILM'
LOW 0
LARGE 0 (.----- rn -1 ,-m<l 1
Zone I - Ash layer

Large t ---+ A T S O
Zone II - Reaction zone
I I
\
\ Zone 111 - Zone of unreacled B
\ I
!
R 0 R 0
LOW @ LARGE 0 I N D I C A T I N G THE DEVELOPMENT
OF THREE ZONES
Figure 1. Concentrationprofiles for the homogeneousmodel for various Thiele modulus.
PARTICLE-PELLET OR GRAIN MODEL 2. In these equations, r* corresponds to the reduced quantity rei/
reo at a position y in the pellet. The grain model has five parameters
In this model the solid pellet is visualized as consisting of a k,, D,, D e ~reo, and grain shape. Detailed numerical solutions for
number of small particles or grains. Surrounding these grains are the isothermal case have been published by Calvelo and Smith
macropores through which the gas has to diffuse to reach the var- (1970) and Sohn and Szekely (1972b, 1974). An approximate an-
ious grains. The reaction occurs at the surface of each grain ac- alytical solution for isothermal case has been proposed by Evans
cording to the sharp interface model. A product layer will form and Ranade (1980), Sohn and Szekely (1972b, 1974) and Rama-
with time in the outer regions of each grain and these will in turn chandran (1982b). The formulation of the problem as above is for
offer some resistance to diffusion. A detailed analysis of this a spherical grain. A general formulation which assumes various
problem has been presented by Calvelo and Smith (1970) and grain shapes such as spheres, long cylinders and slabs has been
Szekely and Evans (1971a, b, c). The rate of reaction per unit vol- presented by Sohn and Szekely (1972b, 1974). The grain shape
ume of the pellet on the basis of the grain model assuming spherical factor may either have a physical basis (as when pellets are formed
grains is: by compaction) or it could be an empirical model parameter. There
is a close analogy between the grain shape factor and the order of
3(1 - 6 ) - - A
(rei)z reaction with respect to solid in the volume reaction model for the
TA =
1+-
re0
ks re0
rciks
Dt?C
( 1-- :;;) (16) situation when D,c
tively.
a.The values of n of 0,0.5 and 0.67 cor-
-+
respond to grain shapes of flat plate, cylinder and sphere respec-
For nonisothermal systems, numerical solutions have been ob-

where tained by Calvelo and Smith (1970) based on pseudo-steady state
E = porosity of the pellet
approximation ( N 3 = 0 in Eq. 18).A complete transient problem
rct = position of the reaction interface within the grain at a for this case has been solved by Sampath et al. (1975). The transient
given time model is necessary for predicting the temperature profiles in the
pellet. A comparison of the temperature profiles obtained by the
and pseudosteady-state model and the transient model was done by
Sampath et al. (1975). In the pseudosteady-state model the maxi-
re,, = initial radius of the grain mum temperature is attained at 7 = 0 while in the transient model
The model equations for the particle-pellet model are then as fol- it is attained after the lapse of a certain length of time. Another
lows: interesting result is that the pseudosteady-state model predicts that
the maximum temperature is always attained in the center of the
V2a = cbf(r*)2R1 (17) pellet. The transient model predicts that the maximum terlipera-
and
Parameter Notation Definition

where R l is defined as
3(1 - c,)k,(T,)R'
I. Thiele Modulus 41
rd)v
2. Particle Biot Number Ui, Dec l ~ d ~ ( T ~ ) l
3. Heat Accumulation Parameter N:, IlRR2MBCy,ks(Tu)Ag/(kprCo)
1. Dimensionless Time r1 vstMBA,k,(T,)/(p,rc,)
5 . Additional Parameters, Bi,v, BiH, /j'and y; the definitions of thew
Same as in Table 1
The dimensionless parameters 41, 7 1 and Bi, are defined in Table
Page 884 November, 1982 AlChE Journal (Vol. 28, No. 6 )

3. MODELSTO ACCOUNT FOR STRUCTURAL CHANCFS DUETO
TABLE M0BR.d Volume Roactlon M0d.b
REACTION
In these models the porosity is assumed to change with local solid
conversion according to some proposed empirical equation. A
Model Name/TvDe
~~~~
References commonly used relation is a linear increase or decrease of porosity
1. Modified Volume Wen (1968);Ishida and Wen (1968);Calvelo and with conversion,
Reaction Model Cunningham (1970);Gidaspow (1972)
2. Modified Hartman and Coughlin (1974,1976);Ramachandran = 0 + q ( 1 - B/B,) (21)
Particle- and Smith (1977b);Georgakis et al. (1979);Ranade where to is the initial porosity and a1 is a constant. Corresponding
Pellet Model and Evans (1980);Ranade and Harrison (1981); to the porosity change the intraparticle diffusivity is assumed to
Garza-Garza and Dudukovic (1981)
3. Single-Pore Petersen (1957);Szekely and Evans (1970); change, and this again is represented by empirical or theoretical
Models Ramachandran and Smith (1977a);Chrostowski and models. For example, if the random pore model of Wakao and
Georgakis (1978);Ulrichson and Mahoney (1980);Lee Smith (1962) is used to describe the intraparticle diffusion, then
(1980) the change in D, can be represented as
4. Distributed Pore- Hashimoto and Silveston (1973a, 1973b);Simons and
Size Models Rawlins (1980);Bhatia and Perlmutter (1980, 1981a,
1981b);Gavalas (1980);Christman and Edgar (1980)
where D, is the value of intraparticle diffusivity at time zero.
The model equations are then solved with varying diffusivity
rather than with constant diffusivity. Other forms of equations of
ture occurs initially in a region close to the surface of the pellet variation of D, have been proposed by Wen (1968),Calvelo and
which then moves towards the center as time progresses. Cunningham (1970),and Fan et al. (1977),and these are all em-
The grain model is particularly useful for the case where the pirical in nature. For instance, the relation used by Fan et al. (1977)
pellet is formed by compaction of particles of very fine size. The is of the following form
grain size has a clear physical meaning for this case. When the
pellets are not formed in this manner the radius for a sphericalgrain
may be approximated as 3/(S,p ). The grain model has been a p
plied to a number of systems suci as the reduction of NiO (Szekely
and Evans, 1971b;Szekely and Lin, 1976;Evans et al., 1976),and
the hydrofluorination of U 0 2 (Costa and Smith, 1971).
In a recent study, Gibson and Harrison (1980)have applied this where a2 and a3 are empirical constants. Computation using this
model to the reaction of H2S with ZnO. In this study all the model form indicated a sigmoid conversion-time behaviour for certain
parameters were either determined or predicted independently. ranges of parameters which is not generally predicted by a model
(For example, k, was obtiined from the rate data obtained with with constant D,. However it may be noted here that such an a p
finely divided powders; the effectivediffusivitywas predicted from proach is purely empirical and does not lead to a detailed under-
the random pore model of Wakao and Smith, 1962.) The only standing of the effect of structural changes. Also the sigmoid be-
parameter which could not be obtained independently was Da. havior of conversion may be due to nucleation effects as discussed
This was obtained by comparison of the rate data with the model later.
predictions. The r-7 predictions of the grain model were in close
agreement with the theory except in the low temperature range
(soOC) where the reaction ceased before complete conversion was Modified Particle-Pellet Models
reached. Sulfur profiles were measured using electron probe mi-
In these models the radius of the grain rc is assumed to change
croanalysis and were found to agree quite closely with the grain
due to the differences in the molal volume of the products and
model predictions in the low temperature range.
reactants. A schematic representation of this model is shown in -
The grain model for nonlinear kinetics has been presented by
Figure 2. The change in radius of the particle can be easily related
Sohn and Szekely (1973b) for the case of a reaction following
by the following equation:
Langmuir-Hinshelwood rate law.
MODELING OF STRUCTURAL. CHANGES W E TO REACTION where rc is the grain size at time t and r a is the initial grain size
and z, is the ratio of the molal volume of the product to the reactant
A number of structural changes occur in the solid during the defined as
course of reaction and a realistic model for gas-solid reactions
should account for these changes. The main reason to incorporate
these into the model is that existing models cannot predict certain
special behavior patterns which have been observed experi- The parameter z, determines the changes in particle (grain)
mentally. Some of these are: (a) leveling-off of conversion-time data dimensions. If zo < 1, the particles shrink during reaction; if z, >
far below complete conversion; (b) existence of maxima in the 1, swelling occurs; for z , = 1, there is no change in particle size as
rate-time data especially for gasification reactions; and (c) strong reaction progresses.
dependence of the reactivity of the solid on pore-size distribu- The porosity changes in the pellet can be related to changes in
tion. the particle size by the following equation:
The main structural changes are due to chemical reaction and
sintering. This section analyzes the effect of changes due to reaction
while effects of sintering will be presented in the next section. The
changes due to reaction are mainly caused by the differences in The change in the diffusivity can be assumed to be proportional
the molal volume of the reactant and the product. This causes a to 2 and this can be incorporated in the model equations. An il-
change in the porosity of the solid and in the effective diffusivity. lustrative result for an %thermal system is shown in Figure 3. The
A number of attempts have been made to model these effects, and conversion levels off to an asymptotic value (which is less than
Table 3 presents a summary of most of these. A comparison of some 100%)for some cases when z, > 1. Such incomplete conversions
of these models has been presented by Lindner and Simonsson have been experimentally observed for reaction of SO2 with
(1981). A discussion of the main classes of models to account for limestone (Hartman and Coughlin, 1974, 1976) and in the hy-
structural changes due to reaction is presented below. drofluorinationof U q (Costa and Smith, 1971). The models which
AChE Journal (Vol. 28, No. 6 ) November, 1982 P m 885

COMPLETELY REACTED
rci - RADIUS OF UNREACTED

CORE
rG -RADIUS OF GRAIN AFTER
SOME REACT ION
rGo - INITIAL GRAIN SIZE
SCHEMATIC OF
A PARTIALLY REACTED GRAIN
SCHEMATIC OF
A PARTIALLY REACTED P E L L E T
Figure 2. Schematic representation of the changing grain size model.
incorporate structural changes (particle size change in this case) 1

are capable of predicting this phenomenon as can be seen from
Figure 3. Incomplete conversion occurs when the porosity at the
surface of the pellet becomes zero (pore closure phenomenon). An -__.
1 1
analytical equation can be derived from the dimensionless time 2, -1 (1- o ) 2 / 31.1- (1 - &)'/3
required to reach pore closure (Georgakis et al., 1979):
Also the maximum conversion of the solid is given by the following

equation when z, > 1 (Ramachandran and Smith, 1977a):
From Eq. 28 it is also seen that if eo is greater than ( z , - l)/z,,

then complete conversion of the solid is possible; otherwise the
phenomenon of pore closure will be observed. The modified par-
ticle-pellet model has been used for the interpretation of experi-
mental data on hydrofluorination of UOz (Ramachandran and
Smith, 197710) and sulfation of limestone (Georgakis et al., 1979).
Both these systems have z , values around three and hence exhibit
the phenomenon of pore closure.
A variable property grain model has been proposed by Ranade
and Harrison (1981) in a recent paper which is similar in concept
to the model discussed above. In this model, the grain radius is as-
sumed to vary under the combined influence of chemical reaction
and sintering. Significant improvements in the match with the
experimental data of the reaction of hydrogen sulfide with zinc
oxide as compared to the constant property grain model have been
achieved.
0 0.2 0.4 0.6 0.8
DIMENSIONLESS T I M E , 7; Single-Pore Models
Figure 3. Effect of moial volume ratio parameter z. on conversion based on
changing grain-size model (Ramachandran and Smith, 1977b). The single pore models approach the problem in an entirely

ENARCED FORE PORE CLOSURE
SOLID 8
ASSOCIATED
W I T H THE
PORE
( b ) AT TIME ZERO (bIZv< 1 (C)ZV> 1
AFTER SOME R E A C T I O N
Flgure 4. Schematk representation of the single-poremodel.
different manner. In this case the model focusses attention on the

changes taking place in a single pore which is supposed to be r e p 2r(Ar/7f)kssd
:(s1,
I r, t) = (30)
resentative of the changes taking place in the pellet. The single pore
can be assumed to be cylindrical with a concentric ring of solid B 1 + &l+)
Dee
associated with it. A schematic representation of this is shown in
Figure 4. The effect of structural changes would lead to an open where Ar is the length of a differential element in the pellet and
pore structure as shown in Figure 4b for the case of z, < 1. When the factor 7 f is introduced to account for the fact that not all
z, = 1 the pore geometry does not change with time. Finally for
pores are oriented in the radial direction. A value of 7f= 3 can be
the case of z, > 1,a decrease in pore size results as shown in Figure used for practical purposes as suggested by Johnson and Stewart
4c and large values of z, can lead to pore closure phenomenon. The (1965).
model parameters are: (1)average pore radius; ( 2 )radius of the The average rate of reaction at position r is obtained by inte-
associated solid; ( 3 )effective pore length; (4)effective diffusivity grating Eq. 30 for the entire pore-size distribution. When this av-
through product layer; and (5)reaction rate constant. The first three erage value and an average value for effective diffusivity are in-
parameters can be obtained independently by measuring the corporated into the mass balance we obtain
surface area and the pore volume of the initial pellet. Thus the main
advantage of this model is the small number of independent model
parameters. The model explains the results of sulfation of calcium
oxide satisfactorily (Ramachandran and Smith, 1977a; Chrastowski where 0,and are the average values at position r defined as
and Georgakis, 1978). A disadvantage of the model is that it does
not take into account the intersections of the reaction surfaces as
the various pores present in the pellet react. This effect becomes
important only after some appreciable conversion has occurred. and
Also the model does not take into account the pore size distribution
which normally exists.
An extension of the single pore model to account for the effect
of bulk flow and reversibility of the reaction has been proposed by
Ulrichson and Mahoney (1980).Such a situation occurs for example Here the pore-size distribution is represented by the function
in the chlorination of MgO and the model amounting for the above f(s1) which is defined such that f (s1,r, t)dsl represents the number
additional complexities fit the data more adequately. of pores of sizes between s1 and s1+ d s l intersecting a unit surface
Lee (1980) has analyzed a parallel pore model and obtained at position r at time t. The pore size distributionitself changes with
analytical solution for x-t behavior by an averaging technique. The time and this was modeled in the work of Christman and Edgar
model is similar in concept to the single pore model and the ana- (1980)by a population balance equation of the type
lytical relations developed by Lee are valid for low and interme-
diate values of Thiele modulus. (34)
Distributed Pore Models Equation 34 is the standard population balance equation which
assumes that there is no net fractional increase in the number of
These models are similar in concept to the single-pore model but pores by birth or death mechanisms; that is, it assumes that
take into account the effect of pore-size distribution. Some of the the total number of pores are conserved. In reaction systems where
recent approaches proposed in the literature are discussed a solid product is formed, there is no mechanism by which two
below. pores may intersect to produce new pores. Hence the birth
Christman and Edgar (1980)applied the single pore model lo- mechanism is absent. (This is unlike the case of gasification systems
cally at a point, r , in the pellet and for a pore of radius s1. This pore where the birth mechanism may also be present.) However pores
is assumed to be covered with an inner concentric product layer can be lost by collisions; that is, if two pores are close enough for
of thickness (s2 - s1). The reaction interface is thus a cylindrical their outer radii of reaction to touch each other, part of the reaction
surface at s2. At time t = 0, s1 = s10 (the initial pore radius) and s2 surface is lost. Incorporation of this will generalize the model
= 0. The values of s 1 and s2 at any time are related by an equation considerably but wiu require a complicated and exhaustive analysis
which is the same as that for a single pore model. of the problem which does not appear to have been made. A key
feature of the approach of Christman and Edgar (1980)is that the
+
s: = 2, sf0 (1 - 2,) s; (29) evolution of the pore structure with time and position can be pre-
The rate of reaction in kmol/s for this single pore is: dicted. Thus we obtainadditional information from the model such

DIFFUSION LENGTH
FOR GAS A
AREA OF UNIFORM
GAS CONCENTRAT ION
(b)
Flgure 5. Development of reactlon surlace accordlng to the random pore model. The shaded area representsunreacted sololid 8. The dotted area represents
the product layer: (a) early stage showlng product layer around each pore; (b) Intermedlatestage, showlng some overlapplng reactionturlacr; and (c) later
stage, showlng full development of product layer and reactlon turlacs for the pertlcular vlew chosen (Bhetla and Perlmutter, 1981a).
as pore-size distribution at various stages of reaction, and this can surface with conversion. The authors then related these to the actual
be matched with experiments for a rational verification of the (or overlapped) surface by an equation similar to that of Avrami
model. (1940).Based on this model Bhatia and Perlmutter (1980)derived
Simons and Rawlins (1980)followed a somewhat different ap- the following equation for the conversion-timebehavior for the
proach to model the effect of pore-size distribution. They postu- case of kinetic control:
lated that each pore in the pellet reaches the surface as a trunk of
a tree. Thus the external surface of the pellet may be assumed to (37)
have a number of holes (or pores) with radii in the range of smln<
s < smaPA number density function can be defined accordingly. where
Thus the number of pores opening at the surface whose radius is
+
between s and s ds is given by 47rR2f(s)ds. The models can then 7 2 is a dimensionless time defined as
k,A,"S,t
1 - ,
be developed for each pore of radius s and then averaged over the
surface pore size distribution to obtain the average flux of the and 9 is a structural parameter defined as
gaseous reactant at the surface. This flux can in turn be related to
the average conversion of solid B. For each pore of radius s, Simons \k = 4nLO(1- c,)/S~ (38)
and Finson (1979) ascribed an associated internal surface area where So and L, are the reaction surface and total length of the
defined by the equation pore per unit pellet volume at time zero. Equation 37 is similar to
other models; for example, if 9 = 0 it reduces to the volume re-
action model with n = 1. Similarly the grain model behavior can
be closely approximated if 9 = 1.Thisalso leads to the conclusion
where KO is a constant called as the pore aspect ratio which is as- that the concept of reaction order with respect to solid is closely
signed a value around five. The pore size distribution functionf(s) related to the structure of the pores of the solid. A relation between
can be described by a suitable mathematical function to complete the order of reaction n and the structural parameter 9 has also
the definition of the problem. This will, of course, depend on the been obtained and this is shown graphically in Figure 6. Thus the
characteristics of the solid. Simons and Rawlins (1980) used the random pore model is able to offer a rational meaning to the re-
following type of relation: action order with respect to the solid.
In subsequent papers Bhatia and Perlmutter (1981a, 1981b)
f(s) = (36) extended this to consider the case of ash diffusion. The equation
21n sz
7rs3 for conversion-timebehavior at any point in the pellet was derived
(SmJ as
The model is mathematically simple to use and has been used
in the interpretation of the initial rate data for the reaction of SO2
and H2S with calcined limestone (Simons and Rawlins, 1980)and
char gasification (Lewis and Simons, 1979).
Recently Bhatia and Perlmutter (1980, 1981a, 198lb) have
proposed a model which takes into account the intersection of the This was incorporated into the diffusion equation for gas A in the
pores as reaction proceeds. The model is called the random pore pellet and the results were compared with the experimental data
model and it assumes that the actual reaction surface of solid B to of Borgwardt (1970) on reaction of S o 2 with limestones The model
be formed by a randomly overlapping set of cylindrical surfaces agreed with experimental data over a wide range of tempera-
of size distribution g(s) where g(s) ds is the total length of cylin- tures.
drical surfaces (per unit volume of space) having radii between s The model thus introduces one parameter 9 to account for the
+
and s ds. A schematic diagram of the model is shown in Figure pore-size distribution of the solid. Bhatia and Perlmutter (1981b)
have derived equations for P for various types of pore-size distri-
5 which depicts the product layer surrounding the pores at different
stages of the reaction. Bhatia and Perlmutter (1980)have derived bution such as bimodal, square, triangular, log-normal, normal,
equations for non-overlapped surface and volume of the reaction etc. For pores of uniform size the value of 9 is shown to be
Page 888 November, 1982 AlChE Journal (Vol. 28, No. 6)

I
ooo 0.5 $0
CORC STRUCTURE C*RUILTCR,
13
u
20
FRACTION OF PORES REYOVEDBY SINTERIMQ,f~

Figure 6. ReIatbnsh@ botweon the sbuctwalparameter d the random pare
model andtha ad.r d readon wlth mp.cttoroHd intho volumreactkm Figure 7. The Mod d pore int-iOnr on the torhdty lactors (Klm
m0d.l ( M b and P.rlrmner, 1980). and smlth,1974).
*=- 1
In (1 - co)
increase in tortuosity was correlated by Kim and Smith (1974)as
a function of the fractionsfp of the pores removed:
while for the general case
where 7j(fP)is the tortuosity factor when a fractionf, of the pores

*= (41) are removed and r f ( o ) is the factor for fp = 0.The correlation for
F(fp)as a function of fp developed on a theoretical basis by Kim
and Smith is shown in Figure 7. A more complicated model for
Further, an analysis of the various pore-size distributions indi- predicting the effect of sintering on the tortuosity factor has been
cates that a uniform pore size leads to the lowest reactivity. For a developed by Chan and Smith (1976).
bimodal pore-size distribution an optimum structure exists for The change in effective diffusivity in the pellet due to the
which reactivity is maximum. This may give useful guidelines in combined effect of chemical reaction and sintering was correlated
pellet preparation by compaction of microporousparticles to yield by Ramachandran and Smith (1977b)by the following equa-
a solid of maximum reactivity. Another interesting feature of this tion:
model is that for certain pore characteristicsa rate maximum may
be observed with increasing conversion. This maximum is ascribed
to the growth of larger reaction surface from initially smaller pores.
At later sta es this is overshadowed by intersections of these (43)
growing surfaces. The model proposed by Bhatia and Perlmutter The only additional information required for the modeling of the
is sufficiently general and predicts much of the observed experi- effect of sintering is the variation of fp with time. Ramachandran
mental behavior. It also tends to unify the various models such as and Smith (1977b)modeled this as a first order process thus:
volume reaction model and grain model in terms of basic structural
parameters.
%
!?
dt
-
= (1 fp)Ajp exp (44)
EFFECTS OF SINERING where Es is the activation energy for sintering and TM is a char-
acteristic temperature corresponding to the onset of sintering.
Many noncatalyfic reactions are accompanied by large tem- Generally TM will correspond to the Tamman temperature which
perature rise in the pellet and this can cause sintering of the pellet. is approximately half the melting point of the solid. The model
Sintering causes a decrease in the effective diffusivity of the pellet parameters introduced to account for the effect of sinteringare Afp,
and this must be accounted for in the modeling of processes where E, and TM and these can probably be estimated by independent
sintering is likely to be important. Evans et al. (1973)accounted sintering measurements on the solid, i.e., by measuring the porosity
for this by assuming an exponential decay in the effective diffu- and the effective diffusivity for various sintering times.
sivity with reaction time. An illustrative result of the effect of sintering on the conver-
Ramachandran and Smith (197%) have accounted for the effect sion-time behavior of an isothermal system is shown in Figure 3
of sintering on the rate of reaction based on the grain model. The for the case of z, = 0.5 (asdotted line). In this case normally pore
basis for this modeling was an earlier study by Kim and Smith closure is not expected due to chemical reaction as the parameter
(1974)who measured the effective diffusivity in nickel oxide pellets z, is less than unity. However the pore closure occurs due to the
during various stages of reduction. Both sintered and unsintered effect of sintering and results in incomplete conversion of the solid
pellets were used in their study. These authors proposed a model as shown in Figure 3.
for decrease in 0,due to sintering. According to this model, the Ranade and Harrison (1979,1981)approached the problem of
primary influence of sintering is to decrease the pore intercon- sintering effects in a different manner. In this model the sintering
nections. This causes a decrease in porosity of the pellet and a process is visualized to result in the combination of adjacent gr*,
drastic increase in the tortuosity factor 7fof the pellet. This increase the large number of small grains are assumed to be gradually re-
in 7 j is not only due to the decrease in the porosity of the pellet but placed by fewer of large radius. The change in grain size is related
also due to removal of some pore interconnections. The fractional to specific surfacearea of the grains which is described as a function

of time by a rate law proposed by Nicholson (1965)(first-order by these authors lumps the diffusing molecules into three species:
decay in surface area). The moderhas been applied to the experi- tar, gases and inerts. The concentration profiles of these speciesare
mental results on the reaction of hydrogen sulfide with zinc then evaluated by describing the process by ternary diffusion and
oxide. viscous flow in Conjunction with a simple pore model for coal. The
model has been compared with some limited experimental data
on subbituminous coal.
DECOMPOSmON REACTIONS
MISCELLANEOUS MODELS
Decomposition reactions are of prime importance in many o p
erations such as calcination, preparation of high surface area metal
oxides, etc. These reactions may be represented as:
-
Crackling Core Model
44 G(s) + Pk) (ii) Park and Levenspiel (1975)proposed the wcalled cracklingcore
model to account for the sigmoidalbehavior of r vs. t plots in some
The decomposition reactions are modeled generally in terms of
-
observed systems. In this model the reaction is assumed to occur
the sharp interface model and the rate of decompition is assumed in two steps:
to be controlled by simultaneous transfer of mass away from and
Physical change
heat to the interface which separates the reactant and decomposed (1) A (nonporous) A (porous)
solid (Narsimhan, 1961;Hills, 1968;Campbell eta]., 1970). At the (v)
decompositioninterface, the concentration of the gaseous product (2) A (porous) product
+
(Vi)
P is assumed to be determined by the equilibrium of reaction (Eq. The first stage is a physical transformation of A from a nonporous
ii). structure to a more reactive porous structun vhich then undergoes
The model equations and results for conversion and temperature reaction according to the second step. The d e l can also be used
of the decomposing solid as a function of time have been well to represent a situation where the reaction of solid occurs in two
documented in the literature. In a recent paper Mu and Perlmutter consecutive steps. For example, in the reduction of hematite the
(1980) modified these models to take into account the effect of steps are
changes in pellet size arising from possible density differences
between reactant and product. Further in this work the effective +Hz +HZ +H2
Fe304 --+Fez03 +FeO Fe (vii)
diffusion coefficient of the product gas in the product layer (Dee)
was assumed to vary with the decomposition temperature T.The The first step results in the formation of a porous structure which
specific relationship of Dec vs. T depends on the diffusion mech- undergoes further reaction. Similar sequence occurs in the re-
anism prevailing in the solid pores. If the diffusion mechanism is duction by hydrogen of manganese oxide (De Bruijn et al.,
of the Knudsen type, D e is~proportional to while for ordinary 1980).
molecular diffusion the exponent of T is in the range of 1.5to 2.0.
The computational results of Mu and Perlmutter (1980) showed (viii)
that the effects due to changes in particle dimensions were more
significant than those due to the variation of Dec with temperature. The crackling core model has been successfully used in the inter-
In another recent study, Prasannan et al. (1982) incorporated the pretation of experimental data in such systems. Although the model
effects of nonequimolal diffusion of the product gas on the rate of can predict the qualitative features of these systems reasonably
decomposition. The effects are not important if the diffusion is in well, it is only an approximation and does not consider the basic
the Knudsen regime, while if the process is controlled by bulk mechanisms responsible for the sigmoidal behavior: One such
diffusion the analysis of the problem should include these ef- effect could be the nucleation effect which is discussed below.
fects.
Gokhale et al. (1975) reported some interesting observations on
the disproportionationof potassium benzoate to terephthalate (used Nucleation Models
-
in the manufacture of polyester intermediates) and benzene. Nucleation effects are often significant in some gas-solid reac-
2 CsH5COOK C&( COOK)z C& (iii) tions such as reduction of metallic oxides. Typical r-t behavior at
low temperatures for these systems shows three periods: (a) in-
The reaction with these pellets always started at the center and duction period; (b) acceleratory period; and (c) decaying period.
expanded radially outward. A delayed diffusion model has been These three stages arise from the following succession of events:
proposed by Kulkarni and Doraiswamy (1980) to explain these (1)formation of nuclei of the metallic phase at localized sites on
unusual observations. In this model the reaction is assumed to take the oxide surface; (2) growth of these nuclei; and (3)overlap of the
place in two steps with COZ being formed as an intermediate in growing nuclei and a decrease in the metal/oxide interface. The
the first step: The second step requires a critical partial pressure length of the induction period is most intimately connected with
of COZ which is first reached at the point of symmetry in the pellet. the rate of formation of nuclei. If the nuclei are formed fast (as is
This explains the initiation of the reaction (second step) at the the case at high reaction temperature or when the surface-to-vol-
center. An analytical solution for the rate of movement of the re- ume ratio is small) the entire surface gets covered rapidly with the
action interface radially outwards has been given by the same metallic phase and a metal/oxide interface develops. The reaction
authors. then proceeds in a topochemical manner. On the other hand, if the
An industrially important class of decompositionreactions occurs rate of nucleation is not very large, the metal/oxide interface will
in coal devolatilization or pyrolysis. These may be represented be irregular. The mathematical modeling of product layer diffusion
as
-
and chemical reaction for these two distinct situations is, of course,
Coal Residues + Volatiles (i.1
quite different. Some aspects of these have been discussed by Sohn
(1978) and Rao (1979). Formation and growth of nuclei in the Hz
although the actual process is very complex involvingthe formation reduction of wustite has been observed by El-Rahaiby and Rao
of several intermediates as pointed out in a review by Anthony and (1979).
Howard (1976).The volatiles which are formed during heating of A rate equation has been developed by Avrami (1939, 1940,
coal diffuses outward and as some of these are highly reactive free 1941) for the r-t relationship in the absence of pore diffusion:
radicals they undergo secondary reactions in the pores of the solid.
A model for intraparticle mass transfer effects in coal pyrolysis has In - = CtN (45)
(l-tr)
been proposed by Gavalas and Wilks (1980).Although the detailed
chemistry of the process is not well understood, the model proposed where C and N are constants and N depends on the rate of for-

mation of nuclei. A more general numerical solution of the problem Guzman and Wolf (1979) also analyzed this problem assuming
has been presented by Ruckenstein and Vavanellos (1975) which a nonuniform distribution of catalyst within the pellet. The theo-
includes both isothermal as well as nonisothermal systems. The retical analysis included two models of the pore structure and
classical Avrami Model has been modified by Bhatia and Perl- uniform, linear and parabolic activity profiles for the distribution
mutter (1979) to include the initial volume of the nucleus and a of the catalyst. The results of the analysis revealed that in gas-solid
population balance approach is used to model the solid-phase re- reactions catalyzed by solids, nonuniform catalyst distribution is
actions in terms of nuclei of the product phase dispersed in the detrimental to solid conversion and efforts should be made to attain
reactant matrix. A conversion-time relationship is obtained by the a uniform catalyst distribution.
solution of a set of moment equations derived from the population Chloromethylation of silicon (or ferrosilicon) to form methyl
balance. A model of nucleus impingement at low conversions is chlorosilanes by copper catalyst is an industrially important system.
formulated and applied to published data on the oxidation of cu- A fluidized bed technology developed by the National Chemical
prous iodide. Laboratory is in operation (Doraiswamy et al., 1980). Even though
The effect of pore diffusion on systems following nucleation systematic modeling of this complex solid catalyzed gas-solid re-
kinetics of the type given by Eq. 45 has been analyzed by Sohn action has not so far been attempted, Voorhoeve (1967) has sug-
(1978). Approximate analytical solutions have also been presented gested a physical picture of the reaction mechanism which is given
assuming the law of addition of diffusion and reaction times. in Figure 8. The reaction is catalyzed by v-Cu3Si, which is formed
during the initial stages of the reaction.
This initial solid-solid reaction was found to be autocatalytic by
Solid Catalyzed Gas-Solld Reactions
Tamhankar et al. (1981a). Silicon from the 7-phase then reacts with
A number of gas-solid reactions are catalyzed by solids. Some methylchloride and leaves the lattice as gaseous product. The va-
industrially important examples are: (1) hydrogasification of coal cant lattice site is occupied by silicon diffusing from the silicon
and coal char catalyzed by alkali-metal or calcium compounds; and phase to the g-phase. The silicon conversion is limited by the
(2) formation of chlorosilanes(used in the manufacture of silicones) blocking of the silicon surface by the g-phase, whereby the rate of
by reaction of methyl chloride and silicon catalyzed by metallic solid state diffusion of silicon to the copper-rich surface becomes
copper. the rate limiting step. The above physical model can form the
Detailed mathematical analysis is not available for these systems. starting point for a detailed mathematical modeling of the pro-
Lee (1979) proposed a simple model for catalyzed gas-solid reac- cess.
tions on the assumption that the catalyst is present as an external
layer surrounding the solid reactant. In this model the reactant gas
is assumed to diffuse through this catalyst layer and the gasification GASIFICATION REACTIONS
of the solid reactant takes place at the catalyst/solid reactant in-
terface. As conversion proceeds, the thickness of the catalyst layer In gasification type of reactions only gaseous products are
increases causing the effectiveness factor to decrease. This model formed. Examples of such reactions are found in the combustion
is a simplified picture and more complete modeling along with of carbon, formation of metal carbonyls, etc. The analysis of such
systematic experiments is necessary. reactions for nonporous particles can be approached using the
CH3CI
I METHYL CHLOROSILANE
IN I T I AL STAGE LATER STAGES
SURFACE OF SOLID COVERED

BY T\-PHASE
m - t\-Cu3Si PHASE
hKy - SILICON PHASE
- - DIRECTION OF Si DIFFUSION
Figure 8. A physical model for reaction of silicon with methyl chloride (Voorhoeve, 1967).

shrinking core model. This has been attempted by Szekely et al. I I I 1
(1976)and Levenspiel (1973).The main feature of the analysis of
gasification reactions on the basis of the sharp interface model is
that no product layer exists, and hence diffusion through the ash
layer which exists in the general type of reactions (Eq. i) is not
present.
For porous particles the change in surface area with reaction
must be considered. The gasification reactions provide a case of
severe structural changes as no solid product is formed during the
reaction. In some cases gasification may only be partial (chars to
activated carbon) but even here the structural changes are very
significant. MDDEL PREDICTIONS OF
0.41
One of the first models to explain this problem was proposed by GAVALAS ( I 9 8 0 I
Petersen (1957)who assumed a random distribution of uniform-
EXPERIMENTAL DATA OF
sized pores. A method was presented for the analysis of this type o.3t/
MAWJAN *? 01 ((9781
of reaction where the kinetic expression is linear in concentration
of the reacting gas and where appreciable concentration gradients 0.21 I I I I
0.5 1.0 2.0
are established in the pore system as a result of intraparticle dif-
fusion. t / to.5
The model of Petersen leads to the following equation for the Flgure 9. A comparlsonof the predlcllonsof randomcaplllary model wlth data
gasification reaction of a porous pellet in the absence of intrapar- on gaslflcatlon of char.
ticle gradients. (Szekely et al., 1976)
In a recent study Gavalas (1980)has proposed a random capillary

model for describing the porous medium. In this model the porous
G -1 solid is described by a single probability density function p ( s ) which
is related but not identical to the customary pore size distribution.
For low porosity this function is defined as follows:
where G is the solution of the cubic equation

4 Based on the probability distribution two structural parameters are
= o~
-C~C~-G+ (47) defined
27
and spa and to is the initial pore radius and porosity respec-
tively.
Hashimoto and Silveston (1973a, b) modeled the gasification
No = llT p(so)dso (51)
reaction on a microscopic scale by allowing for pore growth, ini-

tiation of new pores and coalescence of adjoining pores. Each of
these leads to change in number of pores of a particular size and where p(s,) is the density distribution based on the initial pore-size
of the pore size density distribution function. These changes were distribution.
modeled on the basis of a population balance equation of the fol- The model of Gavalas (1980) predicts the conversion-time pat-
lowing form: tern in terms of the above two structural parameters:
r ( t ) = 1 - exp [-27r(Nou2t2 + 2Nlut)I (53)
where u is the reaction velocity defined as the rate of change of pore
where B is the rate of introduction of new pores of sizes and D radius with time.
is the rate of removal of such pores by coalescence. Equation 48 A rearranged form of Eq. 53 indicates that a plot of l / t In 1/1
was converted into a set of differential equations by the method - x vs. t is a straight line with a slope equal to 27rN0u2 and an in-
of moments (i.e., multiplying by s and integrating from 0 to s). tercept equal to 47rNlU. Similarly by differentiating Eq. 53 and
The equation was then coupled with a rate model of the form eliminating t the following linear equation is obtained:
(49)
This coupled with suitable equations for the rate of formation of These forms of linear equations are very convenient in corre-
new pores and rate of pore coalescence provides the complete lating x - t or ( d x l d t )vs. r data and have been used by Gavalas
formulation of the problem. (1980)to correlate the data of Mahajan et al. (1978)and Dutta and
The effect of intraparticle diffusion was also considered in a Wen (1977) on gasification of char by oxygen. The model com-
subsequent paper. This could be characterized by a Thiele modulus parison is shown in Figure 9. In the absence of a detailed charac-
R o [ k ~ p ~ S o ~ o / v s D eIfo ]the
1 / 2modulus
. is greater than 10 the terization of the porous structure of char the parameters No and
shrinking core model can be used, while if it is less than 0.1 dif- N1 may be viewed as empirical correlating constants. Equation 53
fusional gradients may be ignored in the model formulation. The also predicts a maximum in the reaction rate when the following
model predicts a maximum in the relative surface area vs solid value of conversion is reached.
conversion behavior. The model needs large numbers of pa-
rameters to characterize the system and many of these can be o b
tained only by matching experimental data with theory.
Simons and Finson (1979) and Simons (1979) have also proposed
a similar model. This model deals with various important quantities
xmax rate = 1 - exp
I (1 31
- - - 27r -
Such a maximum in the rate curve has been observed, for instance,
(55)
such as density of pore intersections, length of pore segments. In by Dutta and Wen (1977)and the model of Gavalas correctly in-
this model the gasification rate is calculated by integrating the rate dicates this behavior.
of gasification per pore tree over the postulated pore-size dis- The random pore model of Bhatia and Perlmutter (1980) can
tribution. (See the section on Single-Pore Models.) also be used to model gasification processes. (For such a system the

molal volume parameter z, approaches zero.) The model predicts generalized approach has been reported at this stage, and in this
a semilog relationshipbetween the group [S/1- r]2and (1- r ) section we shall summarize the approaches followed for specific
and this is in conformity with the data of Hashimoto et al. (1979) systems In some cases the postulated physical models are presented,
on steam activation of char. as these often provide the basis for further mathematical analysis
The kinetics of reaction of coal char with CO, was found to of these problems.
follow a nonlinear Langmuir-Hinshelwood relation by Dutta et Repr-ntative types of complex gas-solid systems with some
al. (1977). Inclusion of this behaviour is necessary for a correct important examples are presented in Table 4. A number of schemes
modeling of the process, especially when significantintraparticle have been analyzed on the basis of the sharp interface model by
gradients exist in the pellet. Wen and Wei (1971) for isothermal systems and Rehmat and
A model for intraparticle diffusion effects in char combustion Saxena (1977) for nonisothermal systems. The application of the
has been proposed by Srinivas and Amundson (1980a) incorpo- zone model for complex reactions has been reported in a recent
rating the effects of three independent reactions paper by Tone and Wen (1981).
c -210 2 co
+ -+
Consecutive Reactlons
co -210 2 co2
+ -+
These are encountered in the reduction of haematite by Hz or
CO. The reaction proceeds through the sequence indicated in Eq.
+
COZ c 2 c o +
(xi)
vii. Hence complex models are necessary to represent this problem.
Such models have been presented, for instance, by Spitzer et al.
Based on the published literature kinetics, reactions ix and xi were (1968)and Tsay et al. (1976), which account for the presence of
found to be predominant. The model predicts the existence of three three interfaces (Haematite/magnetite,magnetite/wiistite, wiis-
steady states under certain conditions. It does not however include tite/iron) in the pellet (Figure 10).Each step of the reaction is as-
the effects of structural changes. In a related study the process of sumed to take place at the sharp interface separating the respective
char gasification was analyzed by Srinivas and Amundson (198Ob). product and reactant. The application of other types of models and
The process was assumed to consist of four independent reactions, inclusion of structuralchanges appear to be an interesting possible
namely, steam gasification, carbon dioxide gasification, hydrc- future development in this area.
gasification and water gas shift reaction. A parametric study of the
various factors affecting the gasification process has been Presented Reaction of Two Gases
by these authors.
Reaction of two gases with the same solid is encountered in a
number of situations. Reduction of iron oreswith a mixture of re-
COMPLEX GAS-SOLID REACTKlN SYSTEMS ducing gases such as CO and Hz appears to be the most important
problem belonging to this class. Here the existence of four gaseous
The analysis in the earlier sections was mainly confined to the components (CO,Hg, COz and HzO) necessitates the use of gen-
case where a simple single step reaction takes place. Many of the eralized Stafan-Maxwell equations to characterize the diffusion
industrially important systems follow complex reaction schemes of the gases in the product layer. Such an approach has been fol-
and modeling of these becomes correspondingly difficult. No lowed by Tsay et al. (1976). The sharp interface model was again
TABLE4. %ME EXAMPLES

OF COMPLEX GAS-SOLID
REACTIONS
Tv~e Scheme Examdes Investigation

1 2 3 4
1. Consecutive Reactions of the Solid 1. Reduction of Haematite Spitzer et al. (1966)

2. Reduction of Manganese Oxide De Bruijin et al. (1980)
+B +B
2. Consecutive Reactions of the Gas 1. Reaction of Iron Sulfide with Oxygen Tseng et al. (1981)
-
A1 + P A C 2
3. Reaction of Two Gases with the Same A1 +B Products 1. Reduction of Iron Ore with a Mixture of Tsay et al. (1976)
Solid
A2 + B - Products
CO and Hz
2. Reaction of UOzFz with a mixture of Fz Anastasia et al. (1971)

and BrFs
4. Reaction of Two Solids with the Same A + B1 + Products 1. Chlorination of Ilmenite Doraiswamy et al. (1959)
Gas
A + BZ - Products 2. Regeneration of Coked Catalyst
3. Sulfation of Dolomite
Ramachandran et al. (1975)
Hartrnan and Coughlin (1976)
4. Reduction of a Mixture of Nickel Oxide Szekely and Hastaoglu (1976)
and Iron Oxide
5. Gaseous product of first reaction 1. Reduction of cobalt sulfide in presence Fahim and Ford (1978)
reacting with a second solid of CaO
2. Oxidation of metal sulfide in presence Bartlett and Huang (1973)

of CaO
6 . Solid product of first reaction reacting
with a second gas
Ai + B - G + P
Products
+
G + A2 U A ~ A ~ -
1. Simultaneousoxidation and sulfation of
cuprous oxide
Bourgeois, Jr. et a1 (1974)
2. Direct removal of HzS(A2)from hot Hasatani and Wen (1977)

synthesis gas (A1 = CO; A2 = HzS, Tamhankar et a1 (1981b)
B = Fe203; G = Fe)
7. Homogeneous heterogeneous reactions 1. Reduction of metal chlorides with Rao (1981)
hydrogen
A, + B, G G + P

REACTION
WUSTITE
INTERCACES FOR
/ CARBUU-4 R E M I D N
I C REKtTION)
ZONE
MAGNETITE C02[WLl , COKE-CIIRBOIY

HAEMATITE
'LETED COKE I
I HYDROGEN IN COKE
A*
HAEMATITE
MAGNETITE COKE REACTS WITH 9

4,b
WUSTITE \ACTION ZONE WHERE CARBON
REACTS W I T H O2
REDUCED IRON Figure 11. Model for simultaneous carbon and hydrogen reaction for coke
oxidation (Ramachandran el al., 1975).
Figure 10. Three interface model for reduction of haematlte (Tsay et ai.,
1976).
used as the basis for this development; the two gases diffuse to this
SO2 + -210 2 + CaO - CaSOd (xiii)
sharp interface and react with the solid at this point. A key feature where Me represents a metallic element. A similar reaction se-
of this work is that multicomponent diffusional models have been quence occurs in the reduction of metal sulfides in the presence
used to describe the diffusion of the reacting and product gases in of CaO
gas shift reaction (CO H2O + -

the ash layer. Croft (1979)has indicated that the possibility of water
+
C02 H2) occurring in the
product layer is likely and hence the effect of this reaction on the
Me, S
H2S
+ H2+
+ CaO -
H2S
CaS
+ x Me
+ H2O
(xiv)
(xv)
conversion-time behavior of the solid needs to be analyzed in
detail. This may be of some importance as the product layer which Sohn and Rajamani (1977)have proposed a model for successive
is composed of reduced iron is an excellent catalyst for this reac- gas-solid reaction of these types. The model can be used to deter-
tion. mine the optimum structure of the pellet and to provide useful
Sohn and Braun (1980) have developed a general model for re- guidelines for the design of experimental investigations of these
action of two gases with the same solid. reactions.
Fahim et al. (1978) proposed a model for metal sulfide reduction
in the presence of CaO. In this model the pellet was assumed to
Reaction of Two Components in the Solid consist of CaO and Me, S grains. The latter were assumed to be
nonporous while the CaO grains were assumed to be porous. Fur-
Reaction of two components present in the solid with the same ther each grain of CaO was assumed to consist of spherical
gas occurs in a number of situations, for example, (a) chlorination non-porous sub-grains. The model equations were solved numer-
of ilmenite, (b) sulfation of dolomite etc. In these cases generally ically and compared with experimental results in a subsequent
one of the solids reacts more rapidly with the gas while the second paper (Fahim and Ford, 1978). In these types of reactions the ki-
component of the solid may not react or may undergo a slow re- netics of both reactions should be studied independently and these
action. These systems are again characterized by significant data should be used in modeling the overall process.
structural changes and future models should include these effects.
An interesting example of reaction of two components of the solid
is encountered in the burning of coke from catalyst particles. Here Solid-Solid Reactions with Gaseous Intermediates
coke oxidation is considered to be composed of two steps: oxidation
of hydrogen which occurs relatively rapidly in the initial time An important class of complex reactions falls in the category of
leaving behind a solid consisting of hydrogen depleted coke which solid-solid reactions proceeding through gaseous intermediates.
gets oxidized subsequently. A model has been proposed for this Reduction of metal oxides with carbon is a major example and the
process by Ramachandran et al. (1975).A schematic representation overall reaction may be represented as
of this model is shown in Figure 11. The model predicts the ex-
perimentally observed rapid initial temperature rise in the pellet Me0 +-
2
' C e M e + ~ C O +l--Cv O ~
+
2
(xvi)
fairly well.
A more complex version of reaction of two solid components of Although the above representation indicates that this is a solid-solid
the pellet occurs in the oxidation roasting of sulfide minerals in the reaction, the actual mechanism is through a gaseous intermediate
presence of lime (a process of likely importance in view of the Cop and can be represented as
emphasis on pollution abatement). Here the gaseous product of the
first reaction reacts with the second solid component. The systems
Me0 + CO + Me + C02 (xvii)
may be schematically represented as c02 +c ?== 2co (xviii)
3
-02
2
+Me, S + SO2 + Me, 0 (xii)
A detailed description of modeling of such reactions has been
described in the book by Szekely et al. (1976).A number of systems

TABLE
5. EXAMPLESOF %LID-%LID REACTIONSPROCEEDING the work in the area of noncatalytic reactions has been relatively
THROUGH GASEOUS INTERMEDIATES sparse. One of the reasons is the mathematical difficulties arising
out of the transient nature of the problem in contrast to the gassolid
Reducing catalysticsystem which attains a true steady state. In the following
Solid Metal Oxide Reference section a Brief summary of the work done in this area is pre-
sented.
Graphite Ferric Oxide Otsuka and Kunii (1969) The instability occumng in gas-solid noncatalytic reactions may
Graphite Ilmenite El-Guindy and Davenport (1970)
Graphite Silica Klinger et al. (1966)
be classified into four major categories.
Carbon Hematite Rao (1971, 1974);Sohn and Szekely 1. Geometric instability
(1973a) 2. Thermal instability
Carbon Magnesium Oxide Fruehan and Martonite (1976) 3. Transition in rate-controlling steps
Carbon Iron Oxide Fruehan (1977)
Carbon Manganese Oxide Kor (1978) 4. Oscillatory instability
Carbon Stannic Oxide Padilla and Sohn (1979)
Carbon Ilmenite Wouterlood (1979) Geometric lnstablllty
Carbon Chromium (111) Vodopyanov et al. (1979)
Oxide Geometric instability arises due to certain irregularities (such
Carbon Zinc Oxide Jalan and Rao (1977) as thermocouples, void portions, etc.) in the pellet. When the rate
Carbon Titanium Oxide Barin and Schuler (1980) of reaction decreases as the surface area reduces, the system is
Carbon Zirconium Oxide Biceroglu and Gauvin (1980) stable. In this case the effect of any surface irregularity is not felt
and the reaction interface maintains the same initial shape
(spherical in case of spherical pellet). If the reverse is true (rate
of importance falling in this class have been investigated in recent increases with decreasing area) then the unevenness in the surface
years and most d these have been s u m m d in Table 5. The rate propagates and results in geometric instability. This was first
controlling step is often found to be reaction of carbon with C02 pointed out by Cannon and Denbigh (1957).For isothermal first
(for example, Kor, 1978).Two complexitiesarise in the modeling order systems the analysis of Nelson (1969)and Evans (1974)in-
of these systems which could form interesting problems for future dicates that for minor surface irregularity the interface is stable
research: (1) The reaction of carbon with COZ follows nonlinear and geometric instability exists only for large deformations (such
kinetics. For example, Turkdogan et al. (1970)observed the fol- as a cavity inside the pellet, etc.)
lowing type of rate law for oxidation of carbon
Thermal InstabllHles
r, = bcoz
(56) Thermal instabilitiesoccur due to the linear nature of the rate
1 + KPCO
of heat removal and the sigmoidal nature of the heat generation
where k is the rate constant and K is the equilibrium constant for curves. The point of intersection of these curves represents the
chemisorption of CO on the carbon surface. (2) The metal formed steady state (pseudosteadystate for the case of reacting solids). Thus
during reduction appears to have a catalytic effect on the reaction under certain conditions the pellet can operate in the region of
of carbon (for example, Pa& and Sohn, 1979).The incorporation multiple steady states. Criteria have been proposed in the literature
of these effects on modeling of these systems may be useful. for delineating this region by Shen and Smith (1963) and Aris
A related problem occurs in the chlorination of oxides in the (1967)for sharp interface models and by Calvelo and Smith (1970)
presence of carbon. This is of great importance in Zr and Ti met- for the particle-pellet model. In all these models a pseudo-steady
allurgy and the reaction may be represented as
Me02 + 2C12 + 2C - M e C4 + 2CO
where M e represents metals such as Zr, Ti, Al, etc.
(xix)
state for heat transfer was assumed. Thus the effect of the heat
capacity of the pellet was ignored, thereby rendering the results
less useful practically. A general transient analysis of the problem
may be necessary and some aspects have been investigatedby Wen
The reaction thus involves reaction of two solid componentsand and Wang (1970).
a gas. The products are generally gaseous under the reaction con- The sudden transition in the controllingregime is a consequence
ditions and therefore it can be easily noted that severe structural of thermal instability. This means that the reaction rate accelerates
changes occur during the reaction. The mechanism of the reaction suddenly at some point in time. This has important consequences
is not very clear and there is considerable disagreement among in reactor operation as runaway conditions may occur. Also, a
various workers on this aspect. As a consequencedetailed modeling sudden decrease in the rate could be observed in some cases. I1-
of these types of reactions has not been attempted in the literature. lustrations of this in systems of practical importance do not appear
Biceroglu and Gauvin (1980) have proposed a simplified model to have been published.
where the rate of reaction was assumed to be proportional to the
initial carbon concentration, to the amount of unreacted zirconium Osclllatory Behavlor
dioxide left in the pellet and the chlorine partial pressure with an
exponent of 0.7 to 0.8. Further, the degree of contact between the Oscillatory instability has been observed in some systems due
two solid constituents (Me02 and carbon) may influence the to some complex reaction mechanisms occurring in the solid. This
progress of reaction. Formation of activated atomic chlorine species is typically characterized by oscillations in the weight loss vs. time
on the carbon surface and subsequent diffusion and reaction of data. This has been observed in the reaction of ZnO with HZS at
these on metal oxide surface has also been postulated as a possible high temperatures (Gibson and Harrison, 1980) and has been ex-
reaction mechanism (Barin and Schuler, 1980). A detailed exper- plained as being due to decomposition of ZnO at high temperatures.
imental investigation with a view to identifying the major rate The Zn vapor produced in this decomposition diffusesoutward and
controlling steps may be a prerequisite to a rational model devel- blocks the pore of the catalyst. Another example of this has been
opment of these systems. observed by Lynch and Elliott (1980) in the oxidation of CaS and
has been explained as being due to the formation of CaSO4 in the
pores and its subsequent decomposition.
STABILITY OF GAS-SOLID REACTIONS
Understanding the region of stability of gas-solid noncatalytic COMMENTS ON MODEL EVALUATION

reactions is important in interpretation of laboratory data, model
verification and reactor design. Although considerable work has From the discussions of the previous sections it is Seen that a
been done in analyzing the stability of gas-solid catalytic reactions number of models have been proposed to predict the behavior

of gas-solid noncatalytic reactions. However in order to evaluate centration profile of B in the pellet with model predictions in ad-
the models critically and to apply them to systems of interest, a dition to matching x-t data. This coupled with independent mea-
complete set of experimental data is necessary and this is often surement of the porosity and effective diffusivity provides a
lacking in the literature. In this section we examine the type of data powerful technique in model discrimination and parameter
required and procedures for model evaluation. evaluation for gas-solid systems. Prasannan and Doraiswamy (1982)
An important conclusion that emerges from the previous sections have discussed this technique in detail and applied it to obtain
is that the structure of the solid plays a key role in characterizing sulfur profiles in a pellet of zinc sulfidesubjected to varying degrees
the rate of reaction. Hence a complete characterization of the solid of oxidation. Gibson and Harrison (1980)and Hartman and
must be undertaken in addition to obtaining the usual conver- Coughlin (1976)have also used this technique in earlier studies.
sion-time data. This has not been the practice in some of the earlier Having obtained the data in the above manner, one can then
studiesand hence a critical comparisonof various models using such proceed to model verification and parameter estimation. Here the
data is rather difficult. The porosity and the pore size distribution choice of the model is often important. The most widely used
of the solid turn out to be important parameters in many of the model has been the sharp interface model due to its mathematical
models and hence such measurements are of great importance. In simplicity and this can fit the data easily over a limited range of
addition, the characterization of the solid at various stages of re- operating parameters. However the model may not be satisfactory
action and that of the final product is often important. Such data when extrapolated over a wide range of operating parameter
are very valuable in fitting the models which predict the evolution values, especially for relatively porous pellets. Also, the model does
of pore structure with time. not provide an insight into the phenomena involved in these pro-
A second important quantity that can be determined from in- cesses. The volume reaction models are suitable for porous pellets
dependent experiments is the effective intraparticle diffusivity. but these models account for the structural changes that occur in
These can be measured under nonreacting conditionsby a number the pellet in a purely empirical way. Such changes are important
of techniques. Such measurementsshould again be done for various whenever the molal density of the product is different from that
extents of reaction in order to note any changes in the effective of the solid. The particle-pellet models are suitable when the solid
diffusivity as a function of the conversion of the solid. is formed by compacting grains of fine size. In addition they are
In some systems sintering of the solid reactant or product occurs capable of accounting for the structural changes due to reaction
to a considerable extent. In these cases the effective diffusivities and sintering in a rational manner. When the grains have no real
are significantly altered by the process of sintering; tortuosity physical meaning (that is, when the pellet is not formed by com-
factors of more than 100 have been experimentally reported for paction) an appropriate grain size and shape will have to be as-
highly sintered pellets (Kim and Smith, 1977).For such systems cribed in the model; thus empiricism is introduced to a certain
diffusivity measurements for partly and fully sintered pellets are extent, The single pore models have mathematical simplicity and
necessary. Also the rate of sintering must be determined by mea- very few adjustable parameters but have the disadvantage of in-
suring some property such as surface area or porosity or pore size herent assumption of uniform pore size. Finally the distributed pore
distribution as a function of time at various temperatures. Such data size model and the random pore model incorporate the structural
are needed to predict the effect of sintering on the conversion-time characteristicsin greater detail and hence are useful in fitting data
behaviour. for pellets of differing pore structures. Among the various dis-
Determination of the order of reaction with respect to the gas tributed pore size models, the random pore model is mathemati-
is necessary for a complete model fitting. Such data can easily be cally simple and convenient to use as it does not require solution
obtained by varying the partial pressure of the gas phase species of detailed population balance equations.
(by adding inert diluents, for instance) over a wide range. This A detailed experimental evaluation of the various models is likely
becomes important in cases where the controlling regimes change to be extremely useful and this will require measurement of various
as a function of time. For example, for a relatively nonporous solid properties of the solid as discussed earlier. Finally a comparison
in the initial time the reaction may be kinetically controlled when of various models with a set of complete and well characterized
the observed order will correspond to the true order. At latter stages experimental data will be very useful in pointing out future di-
of reaction diffusional factors can become important due to for- rections of research in the mathematical modeling of these sys-
mation of the product layer. Here the reaction will approach first tems.
order behavior with respect to the gas. To completely characterize
the conversion-time behavior of such a system, an accurate value
of the order of reaction is necessary and this can only be obtained ACKNOWLEDGMENT
by carrying out the reaction over a wide range of bulk gas con-
centrations of the reactant. The order with respect to the solid is The authors wish to thank P. C. Prasannan for help in the
also important if the experimental data are to be fitted with the preparation of this article.
volume reaction models. This is, however, a quantity which cannot
be easily obtained unlike the order with respect to the gas phase
species and may only have to be obtained by curve-fitting the ex- NOTATION
perimental data. Other models do not require a knowledge of the
order with respect to solid and instead require information such = dimensionless concentration of A at a position y in the
as pore size distribution. pellet, A/A,
The intrinsic reaction rate and the kinetic constants can only be = species A or concentration of A at position y
obtained from rate vs time data. Here again data at different = gaseous reactant species
particle sizes and different gas velocities are useful in obtaining = rate constant for sintering (Eq. 44)
realistic model parameters. Also, as most gas-solid reactions are = concentration of A in the bulk gas
accompanied by significant heat effects, care should be taken to = dimensionless concentration of B at a position y in the
maintain isothermal conditions as far as possible. Further, tem- pellet
perature measurements at various positions in the pellet are ex- = concentration of reactant B at position y
tremely useful in assigning the correct temperature to the observed = reactants present in the solid (Table 4)
reaction rate. = concentrations of B in the pellet at time zero
Electron probe micrographs are proving to be of importance in = Biot number for heat transfer, hR/D,
model discrimination and parameter estimation in gas-solid sys- = Biot number for mass transfer, kgR/De
tems. This technique provides a detailed concentration profile of = Biot number for reaction in a particle, Dec/(ksrco)
the solid B as a function of radial position for various extents of = specific heat of solid pellet
reaction. With this additional data it is possible to match the con- = cumulative concentration defined by Eq. 11

= effective diffusivity of A in the pores of the solid B 41 = Thiele modulus for particlepellet model
= effective diffusivity at time zero 7 = dimensionless time for the volume reaction model
= effective diffusivity of gas A in the product layer 71 = dimensionlesstime in the particle-pellet model
= Knudsen diffusion coefficient for gas A 72 = dimensionless time for the random pore model
= molecular diffusion coefficient Tf = tortuosity factor of the pellet
= activation energy of reaction PB ,Pc = density of solid B and C respectively
= activation energy for sintering PP = bulk density of solid
= fraction of pores removed by sintering x = *ition of the zone of completely reacted solid
= pore number distribution in the pellet P = structural parameter in the random pore model
= functional dependency of rate on cumulative con- Y = stoichiometric coefficient for the species indicated in
centration the subscript
= relative tortuosity factor due to sintering
= pore length distribution function LITERATURE CITED
= product solid formed in the reaction
= heat transfer coefficient in the external film Anastasia, L. J.. P. G. Alfredson, and M. J. Steindler, Reaction Model for
= effective thermal conductivity of the pellet the Flourinationof Uranium and Plutonium Compounds in Fluidized-
= gas film mass transfer coefficient Bed Reactors, Ind. Eng. Chem. Proc. Dap. Dew., 10,150 (1971).
= surface reaction rate constant Anthony, D. B., and J. B.Howard, Coal devolatilization and hydrogasif-
= volume reaction rate constant ication, AIChE J., 23,625 (1976).
= total length of the pores per unit volume of the A h , R., Transition between regimes in gas-solid reactions, lnd. Eng.
pellet Chem. Fund.,6,315 (1967).
= order of reaction with respect to gas Ausman,J. M., and C. C. Watson, Mass Transfer in a Catalyst Pellet during
= molecular weight of species B and G respectively regeneration, Chem. Eng. Sci.,17,32.3 (1962).
Avrami, M., Kinetics of Phase Change I. General Theory,J . Chem.Phys.,
= order of reaction with respect to solid 7,1103 (1939).
= structural parameters in the capillary pore model Avrami, M., Kinetics of Phase Change 11. Transformation-timerelations
= transient heat accumulationparameters in the volume for random distributionof nuclei, J . Chem. Phys., 8,212 (1940).
reaction and particle-pellet models respectively Avrami, M., Kinetics of Granulation,Phase Change, and microstructure
= probability density function in the capillary pore Phase Change II1,J. Chem.Phys., 9,177 (1941).
model Barin, I., and W.Schuler, On the kinetics of chlorination of Titanium
= product gas species dioxide in the presence of carbon, Met.Trans., 11B, 199 (1980).
= radial distance in the pellet Bartlett, R. W., and N.H. Huang, Lime-concentrate-pellet roast process
= dimensionless position of the reaction interface in a for treating copper sulfide concentrates,J. Metals., 25,28 (1973).
Beveridge, G. S. G., and P. J.Goldie, Effectivenessfactors and instability
grain, rcdrc, in non catalytic gas-solid reactions. The effect of solid heat capacity,
= rate of reaction of A per unit pellet volume Chem. Eng. Sci.,23,912 (1968).
= radius of the grain at time t at position r Bellman, R., E. S. Lee, and K. M. Wang, Invariant Imbedding and
= position of reaction interface within a grain Solid-Fluid Reactions-I the unreacted Core Model, Chem. Eng. Sci.,
= initial grain radius 32,1051 (1977).
= pellet radius Bhatia,S.K., and D. D. Perlmutter,A population balance approach to the
= universal gas constant modeling of solid phase reactions, AZChE J . , 25,298 (1979).
= pore radius Bhatia, S. K., and D. D. Perlmutter, A random pore model for fluid-solid
= dimensions of the single pore and the product layer reactions: 1 Isothermal, Kinetic Contro1,AlChEJ., 26,379 (1980).
Bhatia, S. K., and D. D. Perlmutter,A random pore model for fluid-solid
formed around it reactions: I1 Diffusion and transport effects, AlChE J , , 27, 247
= surface area of pellet per unit mass (1981a).
= reactive surface area of pellet at any given time (per Bhatia, S. K., and D. D. Perlmutter, The effect of pore structure on
unit volume of pellet) fluid-solid reactions:. Application of the SOp-Iime reaction, AIChE
= shape factor of the pellet J . , 27,226 (1981b).
= reactive surface area at time zero Biceroglu, O., and W. H. Gauvin, The chlorination kinetics of Zirconium
= time elapsed Dioxide in the presence of Carbon, Can. J . Chem. Eng., 58, 357
= temperature of the pellet at position r (1980).
= initial temperature of the pellet Boersma, M. A. M., J. A. M. Spierts and H. S.Van der Baan, Analysis of
a Second order Gas-Solid Reaction in an isothermal particle. Application
= Tamman temperature of the solid to the reduction of oxides with Gaseous Substrates, Chem. Eng. Sci.,
= reaction velocity defined as dsldt 35,1237 (1980).
= volume enclosed by reaction surface (per unit pellet Borgwardt, R. H., Kineticsof the reaction of SO2 with calcined limestone,
volume) at time t E n d r a . Sci. Technof.,4,59 (1970).
= initial volume enclosed by reaction surface Bourgeois,Jr., S. V.,F. R. Groves, Jr., and A. H. Weke, Analysis of fixed
= conversion of the solid reactant B bed sorption: flue gas desulfurization,AIChE J . , 20,93 (1974).
= dimensionless position in the pellet, r / R Bowen, J. H., and C. K. Cheng, Adiffuse interface model for fluid-solid
= ratio of mold volume of the product to the reactant reaction, Chem. Eng. Sci., 24,1829 (1969).
Calvelo, A., and R. E. Cunningham, Kinetics of gas-solid reactions,J .
Catafysk.,17,l (1970).
G r d Later8 Calvelo, A., and J. M.Smith, Intrapellet tramport in gassolid mn catalytic
reactions, Chemeca Proceeding, No. 3 , l (1970).
B = thermicity parameter (Table 1) Campbell, R. R., A. W. D. Hills, and A. Padin, Transport properties of
Y = Arrhenius number (Table 1) porous lime and their influenceon the decompositionof porous compacts
(-AH) = heat of reaction of calcium carbonate, C h m . Eng. Sci.,25,929 (1970).
= pellet porosity Cannon, K. J., and K. J. Denbigh, Studies on gas-solid reactions-I The
= porosity of the product layer oxidation of zinc sulfide, Chem. Eng. Sci., 6,145 (1957).
tG
Carberry, J. J., The catalytic effectiveness factor under non isothermal
= initial porosity of the pellet
2) = dimensionless temperature at position r , TITo
conditions, AIChE J . , 7,350 (1961).
Chan, S. F., and J. M. Smith, Effect of sintering on diffusion in porous
Q, 9 1 = dimensionless local rate of reaction in the volume solids, Indian C h m . Engr., 18,42 (1976).
reaction and particle-pellet models respectively Christman, P. G., and T. F. Edgar, Distributed Pore size model for the
4 = Thiele modulus for volume reaction model sulfationof lime stone, AIChE Meeting, Chicago (1980).
AlChE Journal (Vol. 28,No. 6 ) November, 1982 Page 897

Chrostowski, J. W., and C. Georgakis, Pore plugging model for gas-solid Gokhale, M. V.,A. T. Naik, and L. K. Doraiswamy, Gas-solid Reactions:
reactions, ACS Symp. Ser., 65, (Chem. React. Eng. Houston) 225-37 An unusual observation in the disproportionation of potassium benzoate
(1978). to terephthalate, Chem. Eng. Sd., 30,1409 (1975).
Chu, W. F., and A. Rahmel, The kinetics of reduction of chromium oxide Gower, R. C., A determination of the transport-limitated reaction rate
by hydrogen, Met. Trans. B., 10B, 401 (1979). for a gas-solid reaction forming a porous reaction product, Ph.D. Thesis,
Costa, E. C., and J. M. Smith, Kinetics of noncatalytic, nonisothermal, Lehigh University (1971).
gas-solid reactions: Hydrofluorination of Uranium dioxide, AIChE J., Guzman, G. L., and E. E. Wolf, Effect of non-uniformactivity distribution
17,947 (1971). on catalyzed gas-solid reaction, I G EC Fund., 18,7 (1979).
Croft, V., Diffusion in Mixed-Gas Reduction, Met. Trans., IOB, 121 Hartman, M., and R. W. Coughlin, Reaction of sulfur dioxide with
(1979). lime-stoneand the influence of pore structure, Ind. Eng. Chem. Proc.
De Bruijn, T. J. W., T. H. Soeravidjaya,W. A. Jong, and P. J. Van Den Berg, Des. Deo., 13,248 (1974).
Modeling of the reduction of Manganese oxides with Hydrogen, Chem. Hartman, M., and R. W. Coughlin, Reaction of sulphurdioxide with
Eng. Sd.,35,1591 (1980). limestone and the grain model, AIChE I. , (1976).
22,490
Deb Roy, T., and K. P. Abraham, An analysis of the Pressure Build-up Hasatani, M., and C. Y. Wen, Reactivity of iron oxide sorbents for hy-
Inside a Reacting pellet during Gas-Solid Reactions, Met. Trans., 5, drogen-sulfide removal from a hot-low BTU gas, Final Report, Contact
349 (1974). No. EY-77x-21-0188 for MERC-DOE (1977).
Del Borghi, M., J. C. Dunn, and K. B. Bishoff, A Technique for solution Hehimoto, K., and P. L. Silveston, Gasification: Part I. Isothermal,kinetic
of the Equations for fluid-solid Reactions with Diffusion, Chem. Eng. control model for a solid with a pore size distribution, AIChE J . , 19,
Sd.,31,1065 (1976). 259 (1 973a).
Doraiswamy, L. K., H. C. Bijawat, and M. V. Kunte, Chlorination of il- Hashimoto, K., and P. L. Silveston, Approximation of rate expressions
menite in a fluidized bed, Chem. Eng. Prog., 55,80 (1959). involving pore size distributions, AIChE J., 19,368 (1973b).
Doraiswamy, L. K., S. P. Mukherjee, S. Ramachandran, and A. N. Gokarn, Hashimoto, K., K. Miura, F. Yoshikawa, and I. Imai, Change in pore
Methyl chlorosilanes-NCL Project Report (1980). structure of carbonaceous materials during activation and adsorption
Dudukovic, M. P., and H. S. Lamba, Solution of moving boundary prob- performance of activated carbon, Ind. Eng. Chem. Proc. Des. Deo.,
lems for gas-solid noncatalytic reactions by orthogonal collocation, 16,73 (1979).
Chem. Eng. Sd.,33,303 (1978a). Hayes, P. C., The kinetics of formationof HzO and COz during iron oxide
Dudukovic, M. P., and H. S. Lamba, A zone model for reactions of solid reduction, Met. Trans., 10B, 211 (1979).
particles with strongly adsorbing species, Chem. Eng. Sd., 33, 471 Hills, A. W. D., The mechanism of the thermal decompositionof calcium
(1978b). carbonate, Chem. Eng. Sd.,23,297 (1968).
Dutta, S., C. Y. Wen, and R. J. Belt, Reactivity of Coal and char I. In Ishida, M., and C. Y. Wen, Comparison of kinetic and diffusional models
carbon dioxide atmosphere, Ind. Eng. Chem. Proc. Des. Deo., 16,20 for solid-gas reactions, AIChE J . , 14,311 (1968).
(1977). Jalan, B. P., and Y.K. Rao, The reduction of zinc oxide by carbon: Non-
Dutta, S., and C. Y. Wen, Reactivity of Coal and char 11. In Oxygen Ni- catalyzed reaction, Met. Soc.of CIM, 1 (1977).
trogen Atmosphere, Ind. Eng. Chem. Proc. Des. Den, 16,31 (1977). Johnson, M. L. L., and W. E. Stewart, Pore structure and diffusion in solid
El-Guindy, M. I., and W. G. Davenport, Kinetics and Mechanism of Il- catalysts, J . Catdyskr, 4,248 (1965).
menite Reduction with Graphite, Met. Trans., l , 1729 (1970). Kam, A. Y.,A. N. Hixson, and D. D. Perlmutter, The oxidation of Bitu-
El-Rahaiby, S. K., and Y. K. Rao, The Kineticsof Reduction of Iron oxides minous coal-1 Development of a mathematical model, Chem. Eng. Sd.,
at moderate Temperatures, Met. Trans., 10B, 257 (1979). 31,815 (1976).
Evans, J. W., J. Szekely, W. H. Ray, and Y. K. Chaung, On the optimum Kawahata, M., and P. L. Walker, Jr., Mode of porosity development in
temperature progression for irreversible non-catalytic gas-solid reac- activated anthracite, Proc. 5th carbon Conf., 2,251, Pergamon, London
tions, Chem. Eng. Sd.,26,683 (1973). (1962).
Evans, J. W., Geometric instability in gas-solid reactions, Chem. Eng. Kim, K. K., and J. M. Smith, Diffusion in Nickel-oxide pellets-Effects
Sd.,29,1660 (1974). of sintering and reduction, AIChE J . , 20,670 (1974).
Evans, J. W., S. Song, and C. E. Leon-Sucre, The kinetics of nickel oxide King, Jr., W. E., and Jones, W. S., A solution technique for non-catalytic
reduction by hydrogen, measurements in a fluidized bed and in a diffusion reaction models, Chem. Eng. Sd.,34,1387 (1979).
gravimetric apparatus, Met. Trans., 7B, 55 (1976). Klinger, N., E. L. Stewart, and K. L. Komerek, Reaction between silica
Evans, J. W., and M. G . Ranade, The Grain model for reaction between and graphite, J . Am. Ceram. Soc.,49,369 (1966).
a gas and a porous solid-a refined approximate solution to the equa- Kor, D. J. W., The thermal decomposition of Mn2O3 and the reduction
tions, Chem. Eng. Sd.,35,1261 (1980). of Mn304 by C and CO, Met. Trans.,9B, 307 (1978).
Fahim, M. A,, N. Wakao, and J. D. Ford, Gas-solid reactions I: A grain-cell Kulkarni, B. D., and L. K. Doraiswamy, A delayed diffusion model for
model for complex reactions, Can. J . Chem. Eng.,58,725 (1978). an unusual reaction (Disproportionationof potassium benzoate to ter-
Fahim, M. A., and J. D. Ford, Gas-solid Reactions 11: Reduction of Cobalt ephthalate) involving an expanding reaction zone, Chem. Eng. Sd.,
sulphide in the presence of calcium-oxide,Can. J . Chem. Eng., 56,730 33,817 (1980).
(1978). Lee, H., Effectiveness factors for solid catalyzed gas-solid reactions,
Fan, L. S., K. Miyanami, and L. T. Fan, Transients in Isothermal Fluid- Chem. Eng. Sd.,34,5 (1979).
solid reaction systems-Modeling of the sigmoidal-conversion-time Lee, H., Simple conversion relationship on non-catalytic gas-solid reac-
behavior of a gas-solid reaction, Chem. Eng. J . , 13,13 (1977). tions, Ind. Eng. Chem. Proc. Des. Deo., 19,237 (1980).
Fruehan, R. J., and L. J. Martonite, The rate of reduction of MgO by car- Levenspiel, O., Chemical Reaction Engineering,John Wiley, New York
bon, Met. Trans., 7B, 537 (1976). (1973).
Fruehan, R. J., The rate of reduction of iron oxides by carbon, Met. Lewis, P. F., and G. A. Simons, Char Gasification: Oxidation results,
Trans., SB, 279 (1977). Combustion Sd.Tech.,20,117 (1979).
Garza-Garza, O., and M. P. Dudukovic, Some observations on gas-solid Lindner, B., and D. Simonsson, Comparison of structural models for
noncatalytic reactions with structural changes, Chem. Eng.Sd.,36, gas-solid reactions in porous solids, Chem. Eng. Sd., 36, 1519
1257 (1981). (1981).
Gavalas, G. R., A random capillary model with application to char gas- Luss, D., and N. R. Amundson, Maximum temperature rise in gas-solid
ification at chemically-controlled rates, AIChE ]., 26,577 (1980). reactions, AIChE J . , 15,194 (1969).
Gavalas, G. R.,and K. A. Wilks, Intraparticle mass transfer in coal py- Lynch, D. C., and J. F. Elliot, Analysis of oxidation reaction of CaS,Met.
rolysis, AIChE I.,26,201 (1980). Trans., 11B, 415 (1980).
Georgakis, C., C. W. Chang and J. Szekely, A changing grain size model Mahajan, P. 0.. R. Yarzal, and P. L. Walker, Jr., Unification of coal-char
for gas-solid reactions, Chem. Eng. S d . , 34,1072 (1979). gasification reaction mechanisms, Fuel, 57,643 (1978).
Gibson, J. B., and D. P. Harrison, The reaction between hydrogen sulfide Mantri, V. B., A. N. Gokarn, and L. K. Doraiswamy, Analysis of gas-solid
and spherical pellets of zinc oxide, I & EC, Proc. Des. Deo., 19, 231 reactions: Formulation of a general model, Chem. Eng.Sd., 31,779
(1980). (1976).
Gidaspow, D., Fluid-Poroussolid reaction model with structural changes, Mu, J.. and D. D. Perlmutter, An equilibrium interface model for solid
Intersoc. Energy Convers. Eng. Conf. Proceedings of 7th Conference, decomposition, Chem. Eng. Sd.,35,1645 (1980).
1159-68 (1972). Narsimhan, G., Thermal decomposition of calcium carbonate, Chem.
Gokarn, A. N., and L. K. Doraiswamy, A model for solid-gas reactions, Eng. Sd.,16,7 (1961).
Chem. Eng.Sd.,26,1521 (1971). Natesan, K., and W. 0. Philbrook, Mathematical model for temperature
Gokarn, A. N., and L. K. Doraiswamy, Measurement of diffusion in the variation within a particle undergoing reaction with application to
ash layers in gas-solid reaction, Chem. Eng. Sci., 27,1515 (1972). roasting of zinc-sulphide, Trans. Met. SOC. AIME, 245,1417 (1969).

Nelson, P. A., Geometricalinstability in gas-solid reactions, Chem.Eng. sulfide with calcined limestone, I d . Eng. Chem.Proc. Des. Dm.,19,
sci.,24,919 (1969). 565 (1980).
Nicholson, D., Variation of surfacearea during the thermal decomposition Sohn,H. Y., and M. E. Wadsworth, Procesof extractive m d u r g y eds,
of solids, Trans. Faradoy Soc., 61,990 (1965). H. Y. Sohn and M. E. Wadsworth, Plenum, New York (1979).
Otsuka, K., and D. Kunii, Reduction of powdery ferric oxide mixed with Sohn, H. Y., and J. Szekely, The effect of reaction order in non-catalytic
graphite particles, J. Chem. Eng. Japan.,2,46 (1969). gas-solid reactions, Can.J. Chem. Eng., 50,674 (1972a).
Padilla, R., and H. Y. Sohn, The reduction of stannic oxide with carbon, Sohn, H. Y., and J. Szekely, Structural model for gas-solid reactions with
Met. Trans., 10B, 109 (1979). a moving boundary 111. General dimensionless representation of irre-
Park, J., and 0.Levenspiel, The crackling core model for the multistep versible reaction between a porous solid and a reactant gas, Chem. Eng.
reaction of solid particles, Chem. Eng. Sci., 32,233 (1975). Sci., 27,763 (1972b).
Petersen, E. E., Reaction of porous solids, AIChE J., 3,443 (1957). Sohn, H. Y., and J. Szekely, The effect of intragrain diffusion on the re-
Prasannan, P. C., and L. K. Dorainvamy, Gas-solid reactions: Experimental action between a porous solid and a gas, Chem. Eng. Sci., 29,630
evaluation of the zone model, Chem. Eng. Sci., in press (1982). (1974).
Prasannan, P. C., P. A. Ramachandran, and L. K. Doraiswamy, Demm,; Sohn, H. Y., and J. Szekely, Reactions between solids through gaseous
position reactions: Effect of bulk flow and heat capacity of the solid, intermediates I. Reactions controlled by chemical kinetics, Chem. Eng.
to be published (1982). Sci., 28,1789 (1973a).
Ramachandran, P. A,, M. H. Rashid, and R. Hughes, A model for coke Sohn, H. Y.,and J. Szekely, A structural model for gassolid reactions with
oxidation from catalyst pellets in the initial burning period, Chem. Eng. a moving boundary-IV. Langmuir-Hinshewoodkinetics,,Chem. Eng.
Sd.,30,1391 (1975). Sci., 28,1169 (1973b).
Ramachandran, P. A,, and J. M. Smith, A single-pore model for gas-solid Sohn, H. Y., and K. Rajamani, Successivegas-solid reactions in a porous
noncatalytic reactions, AIChE J., 23,353 (1977a). pellet: Application to the reaction of metal sulfides in the presence of
Ramachandran, P. A,, and J. M. Smith, Effect of sintering and porosity lime, Chem. Eng. Sci., 32,1093 (1977).
changes on the rate of gas-solid reactions, Chem. Eng. J., 14, 137 Sohn, H. Y., The law of additive reaction times in fluid-solid reactions,
(1977b). Met. Trans., 9B, 89 (1978).
Ramachandran, P. A,, and B. D. Kulkarni, Approximate analytical solu- Sohn, H. Y., and R. L. Braun, Simultaneousfluid-solid-reactionsin porous
tions to gas-solid problems, Ind. Eng. Chem. Proc. Des. Den, 19,717 solids: Reaction between one solid and two fluid reactions, Chem. Eng.
(1980). Sci., 35,1625 (1980).
Ramachandran, P. A,, and L. K. Doraiswamy, Analysis of gas-solid re- Sohn, H. Y., and H. J. Sohn, The effect of bulk flow due to volume change
actions with zero order dependency on both gas and solid: concept of in the gas phase on gas-solid reactions: Initially nonporous solids, Ind.
jumping reaction zones, AIChE J., in press (1982). Eng. Chem. Proc. Des. Den, 19,237 (1980).
Ramachandran, P. A,, Analysis of non-catalytic reactions following Spitzer, R. H., F. S. Manning, and W. 0. Philbrook, Simulation of t o p
Langmuir-Hinshelwood kinetics, Chem. Eng. J., 23,223 (1982a). chemical reduction of hematite via intermediate oxides in an isothermal
Ramachandran, P. A,, Analyticalprediction of conversion-timebehaviour countercurrent reactor, Trans. Met. Soc. AIME, 242,618 (1968).
of gas-solid noncatalytic reactions, Chem. Eng. Sci., in press Srinivas, B., and N. R. Amundson, Intraparticleeffects in char combustion:
(1982b). steady-state analysis, Can. J. Chem. Eng., 58,476 (198Oa).
Ranade, P. V., and Harrison, D. P., The grain model applied to solids with Srinivas, B., and N. R. Amundson, A single particle char gasification
varying structural properties, Chem. Eng. Sci., 34,427 (1979). model, AIChE 1..26,487 (1980b).
Ranade, M. G., and J. W. Evans, The reaction between a gas and a solid Szekely, J., and J. W. Evans, Studies in gas-solid reactions: Part I. A
in a nonisothermal packed bed: Simulation and experiment, Ind. Eng. structural model for the reaction of porous oxides with a reducing gas,
Chem. Proc. Des. Deo., 19,118 (1980). Met. Trans., 2,1691 (1971a).
Ranade, P. V., and D. P. Harrison, The variable property grain model Szekely, J., and J. W. Evans, Studies in gas-solid reactions: Part 11. An
applied to the zinc oxide-hydrogen sulfide reaction, Chem. Eng. Sci., experimental study of nickel oxide reduction with hydrogen, Met.
36,1079 (1981). Trans., 2,1699 (1971b).
Rao, Y. K., The kinetics of reduction of hematite by carbon, Met. Trans., Szekely, J., and J. W. Evans, A structural model for gas-solid reactionswith
2,1439 (1971). moving boundary, Chem. Eng. Sci., 25,1091 (1970).
Rao, Y. K., A physicc-chemical model for reactions between particulate Szekely,J., and C. I. Lin, The reduction of nickel oxide discs with carbon
solids occuring through gaseous intermediates-1. Reduction of hae- monoxide, Met. Trans., 7B, 493 (1976).
matite by carbon, Chem. Eng. Sci.,29,1435 (1974). Szekely,J., J. W. Evans, and H. Y. Sohn, Gas-solid Reactions, Academic
Rao, Y. K., Mechanism and the intrinsic rates of reduction of metallic Press (1976).
oxides, Met. Trans., 10B, 243 (1979). Szekely,J., and J. W. Evans, Structural model for gas-solid reactions with
Rao, Y. K., Dual-reaction model for hydrogen reduction of metallic a moving boundary 11. Effect of grain size, porosity and temperature
chlorides, Trans. Instn. Min. Metall., 90, C 47 (1981). on the reaction of porous pellets, Chem. Eng. Sci., 26,1903 (1971~).
Rehmat, A,, and S. 6. Saxena, Multiple nonisothermal noncatalytic gas- Szekely,J., and A. Hastaoglu, Reduction of nickel oxide hematite mixture
solid reactions: Effect of changing particle size, Ind. Eng. Chem. Proc. with hydrogen, Trans. Inst. Mining Met., C78,85 (1976).
Des. Den, 16,502 (1977). Tamhankar, S. S., A. N. Gokarn, and L. K. Doraiswamy, Studies on the
Rehmat, A,, S. C. Saxena, R. Land, and A. A. Jonke, Non catalytic gas-solid industrially important system CuC1-Si: An example of autocatalysisin
reaction with changing particle size: unsteady state heat transfer, Can. solid-solid reactions, Chem. Eng. Sci., 36,1365 (1981a).
J. Chem. Eng., 56,316 (1978). Tamhankar, S. S., M. Hasatani and C. Y. Wen, Kinetic studies in hot gas
Rehmat, A., and S. C. Saxena, Agglomeration of ash in fluidized bed desulfurization:I. Reduction and sulfidation of iron oxide, Chem. Eng.
gasification of coal by steam-oxygen (Air) mixture, Ind. Eng. Chem. Sci., 36,1181 (1981b).
Proc. Des. Den, 19,223 (1980). Tone, S., and C. Y. Wen, Application of the zone model to multiple non-
Ruckenstein, E., and T. Vanvanellos, Kinetics of solid phase reactions, catalytic fluid-solid reactions, Chem. Eng. Sci., 36,273 (1981).
AlChE J., 21,756 (1975). Tsay, Q. T., W. H. Ray, and J. Szekely, The modeling of hematite reactions
Sampath, B. S., and R. Hughes, A review of mathematical models in single with hydrogen plus carbon monoxide mixture, AIChE J., 22, 1064
particle gas-solid non-catalytic reactions, Chem. Engr. (London), 278, (1976).
485 (1973). Tseng, S. C., Tamhankar, S. S., and C. Y. Wen, Kinetic studies on the re-
Sampath, B. S., P. A. Ramachandran, and R. Hughes, Modelling of non- action involved in hot gas desulfurization. 11. Reactions of iron sulfide
catalytic gas-solid reactions-I. Transient analysis of the particle-pellet with oxygen and sulfur dioxide, Chem. Eng. Sci., 36,1287 (1981).
model, Chem. Eng. Sci.,30,125 (1975). Tudose, R. Z., General model of solid-fluid non-catalytic heterogeneous
Shen, J., and J. M. Smith, Diffusional effects in gas-solid reactions, Ind. reactions, Bul. Inst. Politech. I&, 16,241 (1970).
Eng. Chem. Fund., 4,293 (1963). Turkdogan, E. T., R. G. Olsson, and J. V. Vinten, Pore characteristics of
Shettigar, U. R., and R. Hughes, Prediction of transient temperature carbons, Carbon, 8,545 (1970).
distribution in gas-solid reactions, Chem. Eng. J., 3,93 (1972). Turkdogan, E. T., R. G. Olsson, H. A. Wriedt, and L. S. Darken, Calci-
S i mom n , D., Reduction of flouride by reaction with limestone particles nation of lime stone, S M E Trans. AIME, w 4 , 9 (1973).
in a fixed bed, Ind. Eng. Chem. Proc. Des. Deo., 18,288 (1979). Ulrichson, D. L., and D. J. Mahoney, Pore closure model for gas-solid
Simons, G. A,, and M. L. Finson, The structure of coal char: Part I. Pore reactions: the effect of bulk flow and reversibility, Chem. Eng. Sci.,
branching, Comb. Sci. Tech., 19,217 (1979). 35,567 (1980).
Simons, G. A., The structure of coal char: Part 11. Pore combination, Vodopyanov, A. G., A. V. Serebrayakova, and G. N. Kozhevnikov,
Comb. Sci. Tech., 19,227 (1979). Mechanism of the interaction of chromium oxide with carbon, 120.
Sions, G. A., and W. T. Rawlins, Reactionof sulfur dioxide and hydrogen Akad. Nauk. S.S.S.R. Met., 5,11 (1979).

Voorhoeve, R. J. H., Organohalosilanes, precursors to silicones, Elsevier Wen, C. Y., Non-catalytic heterogeneous solid-fluid reaction models,
Publishing Co. (1967). Znd. Eng. Chem., 60,34(1968).
Wakao, N., and J. M. Smith, Diffusion in catalyst pellets,Chem. Eng. Sci., Wouterlood, H. J., The reduction of ilmenite with carbon,J. Chem. Tech.
17,825 (1962). Biotechnol., 29,603 (1979).
Wen, C. Y., and S. C. Wang, Thermal and diffusional effects in nonca- Yagi, J. and J . Szekely, Effects of gas and solids maldistribution on the
talytic solid gas reactions, Znd.Eng. Chem., 62(8),30 (1970). performance of moving bed reactors, reduction of iron oxide pellets with
Wen, C. Y., and L. Y. Wei, Simultaneous non-isothermal non-catalytic hydrogen, AZChE J., 25,800 (1979).
solid-gas reactions, AIChE J., 17,272 (1971).
Wen, C. Y., and N. T. Wu, An analysis of slow reaction in a porous par- Manuscript receiwd October 29,1981: revision receiwd June 16, and accepted June
ticle, AZChE J., 22, 1012 (1976). 29, 1982.
Shear History Effects in the Spinning of

Polymers
During their industrial processing, polymer liquids are usually subjected to JAN MEWlS and
complex flow histories. In principle, these can be taken into account in the con-
stitutive equations. In reality, difficulties arise for complex flows. This fact can GUIDO De CLEYN
be illustrated by the spinning process. Here, experiments are presented in which Department of Chemical Engineering
the upstream section of a spinning device has been changed systematically. A Katholieke Universiteit Leuven
constitutive model and a calculation procedure are suggested that permit an B-3030 Leuven, Belgium
analysis of the upstream effects in the spinning flow. The results indicate that this
analysis predicts well the spinning dynamics in the present experiments, using shear
flow characteristics for the liquids. It is also concluded that the spinning flow can
be altered by upstream changes. The changes in structure seem to be the most
pronounced ones.
SCOPE
The fluid mechanics of polymer processing constitutes an are normally incorporated in a rather arbitrary manner. Ex-
industrial problem that has not been completely solved. Elon- perimentally some shear history effects have been found by
gational flows, in particular transient ones as in spinning, turn Oliver and Ashton (1976), but no quantitative interpretation was
out to be difficult to describe (Denn, 1980). Attempts to model given.
spinning dynamics with rheological characteristics derived from Considering the more general interest of shear history effects
shear flow data, have show systematic errors (Spearot and in complex processing flows, some systematic experiments were
Metzner, 1972; Bankar et al., 1977; Chang and Denn, 1979). performed. They include spinning flow with a variable shear
Some promising results using relaxation time distributions history. As rheological constitutive equation, a structural ki-
should, however, be mentioned (Phan-Thien, 1978). netics model is used (De Cleyn and Mewis 1981) derived from
An additional problem arises in the case of spinning. This type that by Acierno et al. (1976). Model predictions of the spinning
of flow is always preceeded by an upstream flow in a tube and dynamics are used to assess the rheological model and to in-
a nozzle. The resulting initial conditions for the spinning line vestigate the effect of shear history.
CONCLUSIONS AND SIGNIFICANCE

The suggested rheological model makes it possible to take into solutions under investigation is detected. However, the global
account relatively complex shear histories in spinning experi- effect is reduced somewhat by a compensation in the spinning
ments. If the shear rate is not constant throughout the cross flow. The interference becomes more obvious if the resulting
section, as in pipe flow, averaging over the structure distribution structures are calculated. The effect can be associated with
gives better results than averaging over the shear rate. The stress structural changes of a suitable characteristic time scale.
levels at the nozzle can be estimated directly as long as they are The rheological model and the calculation procedure used
not affected by the spinning column. Experimentally some ef- here provide a means to analyze relatively complicated flow
fect of flow history on the spinning dynamics of the polymer patterns and suggest a route to the rational analysis of polymer
processing operations. The model can also be used in numerical
simulations. Further elaboration will be necessary in order to
handle the even more complex situations of industrial pro-
Present address of C De Cleyn. E-hem N. V , Zwijndrecht. Belgium cesses.
0 0 0 1 - 1 5 4 1 - 8 2 - 6 5 2 7 - 0 $ 2 . ~ . 0 The American Institute of Chemical Engineen. 1982.

Modeling of Noncatalytic Gas-Solid Reactions

Uploaded by

Copyright:

Available Formats

Modeling of Noncatalytic Gas-Solid Reactions

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modeling of Noncatalytic Gas-Solid Reactions

Uploaded by

Copyright:

Available Formats

Modeling of Noncatalytic Gas-Solid

CONCLUSIONS AND SIGNIFICANCE

AK=hE Journal (Vol. 28, No. 6) November, 1982 Page 881

AlChE Journal (Vol. 28, No. 6)

Z(T)= 3 1' y2b dy (6)

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 883

Zone I - Ash layer

respond to grain shapes of flat plate, cylinder and sphere respec-

For nonisothermal systems, numerical solutions have been ob-

Parameter Notation Definition

Page 884 November, 1982 AlChE Journal (Vol. 28, No. 6 )

AChE Journal (Vol. 28, No. 6 ) November, 1982 P m 885

rci - RADIUS OF UNREACTED

incorporate structural changes (particle size change in this case) 1

Also the maximum conversion of the solid is given by the following

From Eq. 28 it is also seen that if eo is greater than ( z , - l)/z,,

Page 886 November, 1982 AlChE Journal (Vol. 28, No. 6 )

( b ) AT TIME ZERO (bIZv< 1 (C)ZV> 1

different manner. In this case the model focusses attention on the

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 887

Page 888 November, 1982 AlChE Journal (Vol. 28, No. 6)

FRACTION OF PORES REYOVEDBY SINTERIMQ,f~

where 7j(fP)is the tortuosity factor when a fractionf, of the pores

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 889

Page 890 November, 1982 AlChE Journal (Vol. 28, No. 6)

IN I T I AL STAGE LATER STAGES

SURFACE OF SOLID COVERED

hKy - SILICON PHASE

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 891

In a recent study Gavalas (1980)has proposed a random capillary

where G is the solution of the cubic equation

reaction on a microscopic scale by allowing for pore growth, ini-

Page 892 November, 1982 AlChE Journal (Vol. 28, No. 6)

TABLE4. %ME EXAMPLES

Tv~e Scheme Examdes Investigation

1. Consecutive Reactions of the Solid 1. Reduction of Haematite Spitzer et al. (1966)

2. Reaction of UOzFz with a mixture of Fz Anastasia et al. (1971)

2. Oxidation of metal sulfide in presence Bartlett and Huang (1973)

2. Direct removal of HzS(A2)from hot Hasatani and Wen (1977)

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 893

MAGNETITE C02[WLl , COKE-CIIRBOIY

MAGNETITE COKE REACTS WITH 9

gas shift reaction (CO H2O + -

Page 894 November, 1982 AlChE Journal (Vol. 28, No. 6)

Understanding the region of stability of gas-solid noncatalytic COMMENTS ON MODEL EVALUATION

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 895

Page 896 November, 1982 AlChE Journal (Vol. 28, No. 6)

AlChE Journal (Vol. 28,No. 6 ) November, 1982 Page 897

Page 898 November, 1982 AlChE Journal (Vol. 28, No. 6)

AlChE Journal (Vol. 28, No. 6) November, 1982 Page 899

Shear History Effects in the Spinning of

CONCLUSIONS AND SIGNIFICANCE

Page 900 November, 1982 AlChE Journal (Vol. 28, No. 6 )

You might also like