CHAPTER 2

Fundamentals of Fixed-Bed Reactors

2.1 ANALYSIS OF FIXED-BED REACTORS

Fixed-bed reactors come in all sizes, but their shape is generally restricted
to that of a circular cylinder which is later filled with solid-supported cat-
alyst. Feed enters one end of the reactor and product exits the other end
of the reactor. The component mass balance for a fixed-bed reactor is
!
@CA @CA v @CA @CA
1 vr 1 1 vz
@t @r r @ @z
0 ! 1
1 @ @C 1 @ 2
C @ 2
C
5 DAB @
A A AA
r 1 2 1 1 RA
r @r @r r @2 @z2

where CA is the concentration of component A in moles/m3. We oper-

ate fixed-bed reactors at constant feed and product flow rates; thus,
they operate at steady state, which means @CA =@t 5 0. At high feed
rates, the flow through a fixed-bed reactor, especially one filled with
small solid-supported catalyst pellets or extrudates, behaves in a plug
flow manner. Therefore, vr 5 v 5 0. The reactor design, catalyst size,
and fluid flow rate combine to determine whether the dispersion terms
impact the performance of a fixed-bed reactor; the dispersion terms are
 
1@ @CA 1 @2 CA @ 2 CA
r 1 2 1
r @r @r r @2 @z2
The Peclet number for the process and the various aspect ratios of
the fixed-bed reactor determine the impact of the dispersion terms. The
mass Peclet number quantifies the ratio of bulk mass transport to dif-
fusive mass transport. We define the mass Peclet number as
Lv
Pe 5
DAB
where L is a characteristic length; v is fluid velocity; and DAB the diffusiv-
ity constant for component A in bulk component B. L can be the height
14 Adiabatic Fixed-bed Reactors

of the catalyst mass, identified as Z, or the diameter of the reactor, identi-

fied as D. The product of the directional aspect ratio and the Peclet num-
ber determines the importance of dispersion in that direction. If
Z=Dp  Pe is large, where Dp is the diameter of the solid-supported cat-
alyst, then dispersion in the axial direction; i.e., along the z-axis, is negli-
gible. Generally, Z is large for a fixed-bed reactor and Dp is small; thus,
we can confidently neglect the @2 CA =@z2 dispersion term. With regard to
the axial aspect ratio itself, Carberry1 purports that Z=Dp $ 150 ensures
no axial dispersion in a fixed-bed reactor. In the radial direction, we seek
the opposite outcome because conduction is the only mass transfer mech-
anism in that direction. Thus, we want a small radial Pe number, which is
Dv=DAB . It is heat transfer at the fixed-bed reactor wall that induces
radial conduction and mass transfer. Hence, radial dispersion can only be
neglected for adiabatic reactors. However, Carberry2 writes that radial
aspect ratios R=Dp of 3 to 4 ensures negligible radial dispersion, R being
the reactors radius. Thus, @CA =@r is zero. Chemical engineers generally
neglect dispersion in the azimuthal direction. Therefore, the above com-
ponent mass balance reduces to
dCA
vz 5 RA
dz
where RA represents reactant consumption.
We generally do not characterize fixed-bed reactor operation by its
fluid velocity. Instead, we characterize fixed-bed operation by its volu-
metric flow rate. Rearranging the above equation and multiplying by
the cross-sectional area of the reactor, we obtain
vz A dCA 5 RA A dz
Note that vzA is volumetric flow rate and A dz is dV, the differential
volume of catalyst. V is the fluid volume of the catalyst mass, which is the
true volume of the reactor. However, the CPI generally identifies V as the
weight W of the solid-supported catalyst charged into the reactor shell.
Using W instead of V complicates our analysis for a fixed-bed reactor.
Therefore, we will identify V as the true volume for a fixed-bed reactor.
That volume is

WCatalyst
5 VCatalyst
LBD
 Fundamentals of Fixed-Bed Reactors 15

where WCatalyst is the weight of catalyst charged into the reactor shell
and LBD is the loose bulk density of the catalyst. We use loose bulk
density because we do not shake a commercial reactor while filling it.
We may shake a laboratory reactor tube or shell while filling it, but we
generally do not shake commercial reactors when filling them. Note
that VCatalyst includes fluid volume as well as solid volume. We are
interested in the fluid volume of the catalyst mass, which is

VFluid 5 VCatalyst

where is the void fraction of the solid-supported catalyst.

Thus, the above equation becomes

Q dCA 5 RA dVFluid

Rearranging this last equation gives us

dCA dVFluid
5
RA Q

The boundary conditions for this equation are CA 5 CA,In at V 5 0,

and CA 5 CA,Out at V 5 V. Integrating with these boundary conditions
yields
CA;Out V
dCA dVFluid VFluid
5 5
CA;In RA 0 Q Q

VFluid =Q has units of time, generally in minutes or seconds. We call

VFluid =Q space time and it represents the average time required to tra-
verse a given flow path from the leading edge of the catalyst mass to
the trailing edge of the catalyst mass. We cannot as yet integrate the
left-hand side of the above equation because we have not specified RA
mathematically. To do so, we must know what happens within the cat-
alyst mass during reactor operation.

But, just what does occur within the catalyst mass during reactor
operation? First, fluid enters one of the myriad flow tubes or channels
passing through the catalyst mass. These flow channels form between
catalyst pellets or extrudates. The bulk fluid moves through the cata-
lyst mass via these flow tubes or channels.
16 Adiabatic Fixed-bed Reactors

Second, whenever a fluid flows over a solid or liquid surface, a stag-

nant film forms along that surface. At the surface, fluid velocity in the
direction of bulk flow is zero. At the outer edge of the stagnant film,
fluid velocity is that of the bulk fluid. Within the stagnant film, a
velocity gradient in the direction of bulk fluid flow exists. However,
there is no fluid velocity component normal, i.e., perpendicular to the
solid or liquid surface. In other words, convection does not occur
across the stagnant film. Thus, reactant and product molecules diffuse
across the stagnant film.
Third, most solid-supported catalysts are porous, which greatly
increases their surface area. Increasing surface area leads to an increased
number of catalytic sites available for reaction, thereby increasing cata-
lyst productivity. Reactant molecules migrate along these pores via dif-
fusion. When they encounter an empty catalyst site, they become
product. These product molecules must then diffuse through the pore
network of the solid-supported catalyst, diffuse across the stagnant film
surrounding each catalyst pellet or extrudate, then enter the bulk fluid
to exit the catalyst mass. The concentration difference between a feed
sample and a product sample represents the sum of these mechanisms.
Figure 2.1 presents a schematic of catalysis in a porous solid. The
reactant concentration at the outer boundary of the stagnant film is
the concentration of reactant in the feed, i.e., it is CBF in moles/m3.

Bulk fluid
CBF Pore Catalytic site
CAS CEQ

(CSF CAS)

(CAS CEQ)

(CBF CSF)
Stagnant film
CSF

Figure 2.1 Schematic of adsorption impedances.

 Fundamentals of Fixed-Bed Reactors 17

Reactant molecules move across the stagnant film by molecular diffu-

sion, which we generally model as a linear concentration difference.
That difference is CBF 2 CSF, where CSF is the concentration of reac-
tant at the surface of the catalyst in moles/m3. Reactant then diffuses
from the surface of the catalyst along pores to the catalytic sites
inside the solid. Reactant movement within the pore is also by molec-
ular diffusion, which we model as a linear concentration difference.
Catalytic sites occur along the length of the pore, thus reactant con-
centration changes along the length of the pore. Reactant concentra-
tion at a given catalyst site is CAS in moles/m3. Thus, the
concentration difference to that point in the pore is CSF 2 CAS.
Equilibrium may be established at the catalytic site; equilibrium con-
centration is CEq in moles/m3.
We can write the rate of each of the above described mechanisms as
a concentration difference. The reactant conversion rate is
RRxn 5 kRxn CAS 2 CEq (2.1)
where kRxn is the reaction rate constant at the catalytic site. kRxn has
units of 1/minutes (min) or 1/seconds (s). The rate of reactant move-
ment along the pore is
 
RPD 5 kPD AP =VP CSF 2 CAS (2.2)
where kPD is the pore diffusion mass transfer rate constant (m/s), AP is
the average cross-sectional area of a pore (m2), and VP is average pore
volume (m3). The rate of reactant movement through the stagnant film
surrounding the catalyst pellet is
 
RSFD 5 kSFD SFilm =VFilm CBF 2 CSF (2.3)
where kSFD is the stagnant film mass transfer rate constant (m/s), SFilm
is the surface area (m2), and VFilm is the volume (m3) of the stagnant
film surrounding the catalyst pellet or extrudate.
The only reactant concentrations known with any accuracy are CBF
and CEq. Thus, the mathematical expression for the overall rate of
reactant conversion must be in terms of CBF and CEq. Solving
Eq. (2.1) for CAS yields

RRxn
1 CEq 5 CAS (2.4)
kRxn
18 Adiabatic Fixed-bed Reactors

Solving Eq. (2.2) for CSF gives

RPD
1 CAS 5 CSF (2.5)
kPD AP =VP

Substituting CAS from Eq. (2.4) into Eq. (2.5) yields CSF, thus

RRxn RPD
1 1 CEq 5 CSF (2.6)
kRxn kPD AP =VP

Solving Eq. (2.3) for CSF gives

RSFD
1 CBF 5 CSF (2.7)
kSFD SFilm =VFilm

then substituting CSF from Eq. (2.7) into Eq. (2.6) yields

RRxn RPD RSFD

1 1 1 CEq 5 CBF (2.8)
kRxn kPD AP =VP kSFD SFilm =VFilm

Rearranging Eq. (2.8) expresses the rate of reactant conversion in

terms of CBF and CEq. Thus

RRxn RPD RSFD

1 1 5 CBF 2 CEq
kRxn kPD AP =VP kSFD SFilm =VFilm

By assuming RRxn 5 RPD 5 RSFD 5 R, then rearranging the above

equation, the overall rate of reactant conversion in terms of CBF and
CEq becomes
 
1 1 1
R 1 1 5 CBF 2 CEq (2.9)
kRxn kPD AP =VP kSFD SFilm =VFilm

or
 
1
R5 CBF 2 CEq
1=kRxn 1 1=kPD AP =VP 1 1=kSFD SFilm =VFilm

The overall reaction rate constant is by definition

 
1
kOverall 5
1=kRxn 1 1=kPD AP =VP 1 1=kSFD SFilm =VFilm
 Fundamentals of Fixed-Bed Reactors 19

Inverting kOverall yields

1 1 1 1
5 1 1 (2.10)
kOverall kRxn kPD AP =VP kSFD SFilm =VFilm
We can simplify Eq. (2.10) by combining 1=kRxn and
1=kPD AP =VP . Doing so yields
1 kPD AP =VP 1 kRxn 1
5 1 (2.11)
kOverall kPD AP =VP kRxn kSFD SFilm =VFilm
which reduces to
1 1 1
5 1 (2.12)
kOverall kRxn kSFD SFilm =VFilm
where is defined as
kPD AP =VP
5
kPD AP =VP 1 kRxn
We call the effectiveness factor. The effectiveness factor
accounts for the concentration difference along the pore of a solid-
supported catalyst. If kPD AP =VP ckRxn , i.e., if the fixed-bed reactor
is reaction rate limited, then 5 1 and if kRxn ckPD AP =VP , i.e., if the
fixed-bed reactor is pore diffusion rate limited, then , 1. depends
only upon the pore structure of the solid-supported catalyst and is
readily calculated via a variety of published methods.37

We can now specify RA and substitute it into the component

balance
CA;Out V
dCA dVFluid VFluid
5 5 (2.13)
CA;In RA 0 Q Q

and solve for CA.

2.2 IMPACT OF DIFFUSION

Since the publication of Hougen and Watsons8 classic volume Chemical
Process Principles: Kinetics and Catalysis in 1947, the parameter used for
characterizing flow through processes containing a solid, whether cata-
lytic or noncatalytic, has been space velocity. Space velocity is Q=V , the
inverse of space time V=Q. Note that we have dropped the subscript
20 Adiabatic Fixed-bed Reactors

Fluid on V; henceforth, V means the fluid volume of the catalyst mass,

unless otherwise identified. Thus, space velocity has units of
m3 =s 1
5
m3 s
Q=V has the same units as a first-order reaction rate constant. This
equivalence between an inverse physical time and an inverse chemical
reaction time is generally made when discussing a fixed-bed process.
However, these two inverse times are only equivalent when the process
is reaction rate limited. These two inverse times are not equivalent if
the process is diffusion rate limited.
For a diffusion rate limited process, the rate of reactant consump-
tion is
RDiffusion 5 2 kOverall CBF 2 CEq (2.14)
and for a reaction rate limited process, the rate of reactant consump-
tion is
RReaction 5 2 kRxn CBF 2 CEq (2.15)
Both processes are described by a first-order equation; however,
note that kRxn has dimensions of inverse chemical reaction time, while
kOverall has dimensions of inverse physical time. For a fixed-bed, cata-
lyzed process that is reaction rate limited, we substitute Eq. (2.15) into
Eq. (2.13) to get
  CBF;Out CBF;Out
V dCBF dCBF
5 5 (2.16)
Q Reaction CBF;In RBF CBF;In kRxn CBF 2 CEq

Rearranging, then integrating Eq. (2.16) yields

  
V CBF;Out 2 CEq
kRxn 5 ln (2.17)
Q Reaction CBF;In 2 CEq
For a diffusion rate limited process, we substitute Eq. (2.14) into
Eq. (2.13). Making these substitutions, then integrating yields
  
V CBF;Out 2 CEq
kOverall 5 ln (2.18)
Q Diffusion CBF;In 2 CEq

But, V =QReaction designates the time for a molecule of the bulk fluid,
BF, to react while V =QDiffusion designates the time for a molecule of
 Fundamentals of Fixed-Bed Reactors 21

the bulk fluid to diffuse, then react. Assuming that the right sides of
Eqs. (2.17) and (2.18) are equal gives
   
V V
kRxn 5 kOverall
Q Reaction Q Diffusion
or    
kRxn V V
5
kOverall Q Reaction Q Diffusion
Substituting for 1=kOverall , then performing the necessary algebraic
simplification yields
8 9
>
> >
>
< 1 1 1 = 
kRxn 1  . 1  .  V
Q
> k >
>
: Rxn kPD AP kSFD SFilm >
; Reaction
  V P V Film

5 VQ
8 Diffusion 9
>
> >
>
< kRxn kRxn = 
11  . 1  .  V
Q
>
> A S >
>
: kPD P kSFD Film ; Reaction
  V P V Film

5 VQ
Diffusion
(2.19)
Thus, to obtain the same reactant conversion, the space time of a diffu-
sion rate limited process must be greater than the space time for an equiva-
lent reaction rate limited process. This point is important because,
historically, we upscale and downscale fixed-bed processes using space
velocity Q=V . When developing a new fixed-bed, catalyzed process, we
upscale using laboratory and pilot plant reactor data to design
commercial-sized reactors. When supporting an existing fixed-bed, cata-
lyzed process, we downscale using commercial reactor data to design labo-
ratory and pilot plant reactors. Table 2.1 contains the space velocities for
various sized fixed-bed reactors for an actual process. The solid-supported
catalyst is the same in each of the reactors listed in Table 2.1, as is the
feed. The space velocities of CommercialPlant 2, Catalyst Test Unit
Pilot Plant, and LaboratoryUnit 2 are essentially the same: namely
Q=V 5 10. However, the linear, interstitial fluid velocities for these
22 Adiabatic Fixed-bed Reactors

Table 2.1 Comparison of Space Velocities for a Process Using Fixed-Bed Reactors
Reactor Scale Space Velocity (1/h) Interstitial, Linear Fluid Reynolds Number
Velocity (m/h)

CommercialPlant 1 18 165 92,500

CommercialPlant 2 11 94 52,900
Recycle Pilot Plant 13 4 15
Catalyst Test Unit Pilot Plant 10 0.4 1
LaboratoryUnit 1 60 7 18
LaboratoryUnit 2 10 2 1

reactors are quite different. The Reynolds number for CommercialPlant

2 is 52,900, while the Reynolds number for LaboratoryUnit 2 is one.
The CommercialPlant 2 reactor operates in the turbulent flow regime
while the LaboratoryUnit 2 reactor operates in the laminar flow regime.
These two reactors are comparable only if the process is reaction rate lim-
ited, which it is not. This process is diffusion rate limited; therefore, none
of the laboratory or pilot plant reactors model the commercial plant reac-
tors. If this point is not understood, then a great deal of money will be
wasted doing experiments in the laboratory and pilot plant reactors,
experiments that have no meaning for the commercial plant.

2.3 IMPLICATION OF FLOW REGIMES AND RATE

CONTROLLING MECHANISMS
Commercial plants operate at high volumetric flow rates because their pur-
pose is to produce as much product per unit time as possible. Laboratory
and pilot plant facilities have a different purpose. Their purpose is to pro-
duce quality information that can be used to design a new commercial pro-
cess or to support the operations of an existing commercial facility. We
design commercial facilities to store large volumes of feed and product and
to circulate large quantities of process fluids safely and with minimum
environmental impact. We operate laboratory and pilot plant reactors at
low volumetric flow rates because we want to minimize the volume of feed
stored at the research facility. Another reason we operate laboratory and
pilot plant reactors at low volumetric flow rate is waste disposal: the prod-
uct produced by such reactors cannot be sold. Therefore, it must be dis-
posed, which is expensive. Also, laboratory and pilot plant processes
circulate small quantities of process fluids so that any leak is small, thereby
minimizing the safety and environmental issues arising from the leak.
Finally, if we are developing an entirely new process, we will be unsure of
all the potential hazards it entails. Thus, we will keep all reactive volumes
small to reduce the impact of a runaway or contaminated reaction. In
 Fundamentals of Fixed-Bed Reactors 23

other words, laboratory and pilot plant fixed-bed reactors generally oper-
ate in the laminar flow regime while commercial-sized fixed-bed reactors
operate in the turbulent flow regime.

This flow regime difference can have a dramatic impact upon

the results produced by a laboratory or pilot plant fixed-bed reactor.
In the laminar flow regime, the stagnant film surrounding each solid-
supported catalyst pellet or extrudate will be much thicker than the
stagnant film surrounding solid-supported catalyst in a commercial
reactor operating in the turbulent flow regime.9 The diffusion rate con-
stant, expressed as m/s, is the same in both flow regimes; however, the
time to cross the stagnant film is larger for laminar flow than for tur-
bulent flow since the stagnant film is thicker for laminar flow than for
turbulent flow. Thus, product formation will be slower in the laminar
flow regime than in the turbulent flow regime.

If the laboratory and pilot plant fixed-bed reactors are stagnant

film diffusion rate limited, then kOverall will plot as scatter around an
average value, which may or may not be recognized as kSFD, the stag-
nant film diffusion rate constant. Not being aware of this potentiality
produces highly expensive, inconclusive process support efforts and
catalyst development programs. Years can be spent evaluating differ-
ent catalyst sizes, shapes, and compositions to no avail because all the
data scatters about one value. Unbeknownst to those working on the
project, that point is the stagnant film diffusion rate constant.
Eventually, the project will be canceled due to no conclusive results.

A similar situation can arise for pore diffusion rate limited pro-
cesses. In this situation, the physical structure of the solid support
must be altered to improve catalyst performance, i.e., to increase
kOverall. Changing the chemical composition of the solid-supported cat-
alyst will not alter kOverall. All the data produced by the laboratory
and pilot plant reactors will scatter about one kOverall, which is the
pore diffusion rate constant for the process.

If a solid-supported catalyst is reaction rate limited, then changing

size, shape, or pore structure will not improve catalyst performance. In
this case, the only way to improve catalyst performance is to alter the
chemical composition of the catalyst.

The message is: know the flow regime occurring in a fixed-bed reac-
tor and know the rate controlling step of a solid-supported catalyzed
process before attempting to improve process or catalyst performance.
24 Adiabatic Fixed-bed Reactors

2.4 IDENTIFICATION OF RATE CONTROLLING REGIMES

Consider Eq. (2.12), which is
1 1 1
5 1 (2.12)
kOverall kRxn kSFD SFilm =VFilm
Equation (2.12) is analogous to the resistance present in an electri-
cal circuit. In other words, the overall resistance is the sum of the indi-
vidual resistances present in the circuit. In our case, we are measuring
the resistance to form product molecules. We can determine kOverall
directly by measuring CBF,In and CBF,Out, then calculating kOverall from
the below equation
R 5 kOverall CBF;Out 2 CBF;In
We can determine the value of kRxn from laboratory experiments
that obviate any diffusion effects. What we do not know is kSFD(SFilm/
VFilm). However, we do know that kSFD is constant at a given reactor
operating temperature and pressure. Thus, only SFilm/VFilm responds
to changes in the velocity of the bulk fluid over the surface of the cata-
lyst. Remember SFilm/VFilm is the ratio of the stagnant films surface
area to its volume; therefore, it is inversely proportional to the stag-
nant films thickness. Mathematically
SFilm 1
~
VFilm
where is the thickness of the stagnant film surrounding each solid-
supported catalyst pellet or extrudate. But, stagnant film thickness is
proportional to the Reynolds number of the bulk fluid, i.e.,10
1 1 p
~ p 5 p 5 =Dvz
Rez Dvz =
Thus, as the interstitial, linear velocity of the bulk fluid increases,
decreases. And, as decreases, reactant and product molecules spend
less time traversing the stagnant film surrounding each solid-supported
catalyst pellet or extrudate, which increases kOverall. We can relate to
vz since the fluid density, fluid viscosity, and reactor diameter remain
constant during the velocity change. Thus
SFilm
~ f vz
VFilm
 Fundamentals of Fixed-Bed Reactors 25

f(vz) represents an unspecified function of vz. Equation (2.12) can now

be written as

1 1 1
5 1
kOverall kRxn kSFD  f vz

which has the form of a straight line if we plot 1=kOverall as a function

of 1=f vz . The slope is 1=kSFD and the intercept is 1=kPD .
Figure 2.2 shows a schematic of such a plot. Drawing a horizontal
line through the intercept demarcates film diffusion from pore diffu-
sion and reaction rate. Below that horizontal line, the catalytic process
is limited by a combination of pore diffusion rate and reaction rate.

Figure 2.3 shows actual data for olefin isomerization using a

porous, solid acid catalyst. Figure 2.3 presents 1=kOverall as a function
p
of 1= vz . It represents four fixed-bed laboratory and pilot plant reac-
tors of different sizes. The data presented in Figure 2.3 was collected
over a period of years. Considering the timeline of the project,
R2 5 0.8553 is quite good.
Figure 2.4 presents data collected over a number of years for olefin
metathesis using a metal impregnated, solid-supported catalyst. The
p
data is plotted as 1=kOverall as a function of 1= vz . The correlation is
reasonable considering the time span of the project. Using plots similar
to Figures 2.3 and 2.4, we can determine when a process shifts from

1/kOverall

Resistance to product formation due to stagnant film diffusion rate

Resistance to product formation due to reaction rate and pore diffusion rate

1/f(vz)

Figure 2.2 1/kOverall as a function of 1/f(vz).

26 Adiabatic Fixed-bed Reactors

0.5
0.45
0.4
0.35
0.3
1/k overall

0.25
0.2
0.15 y = 2.8513x + 0.0556
R2 = 0.8553
0.1
0.05
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
(1/v(z))^0.5 in h/cm

Figure 2.3 Olefin isomerization by porous, solid acid.

0.12

0.1 y = 0.9458x + 0.0295

R2 = 0.919
0.08
1/k overall

0.06

0.04 Stagnant film diffusion rate limited

0.02 Pore diffusion + reaction

rate limited
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
(1/v)^0.5 in h/cm

Figure 2.4 1/kOverall as a function of (1/v)^0.5 for Olefin metathesis.

being pore diffusion and reaction rate limited to being stagnant film
diffusion rate limited. The linear correlation for Figure 2.4 is
 
1 1
5 0:946 p 1 0:0295
kOverall vz
 p
At 1= vz 5 0:031, 1=kOverall 5 0:0588. Subtracting 0.0295 from
p
0.0588 gives 0.0293. At this value of 1= vz , the resistance to product
formation is balanced between pore diffusion plus reaction rate resis-
tance and stagnant film diffusion resistance. In other words, the length
 Fundamentals of Fixed-Bed Reactors 27

from the horizontal axis to the horizontal line intersecting the y-axis
equals the length of the line from intersecting horizontal line to the
p
line representing the function. At 1= vz 5 0:032, the process is stag-
p
nant film diffusion rate limited; at 1= vz 5 0:030, the process is pore
diffusion plus reaction rate limited.
We must know which resistance controls product formation before
launching a catalyst or process improvement project. If stagnant film
diffusion rate controls product formation, then the only way to
increase product formation is to increase the interstitial fluid velocity
through the catalyst mass. If pore diffusion plus reaction rate controls
product formation, then we must first decide which resistance is the
major component of their sum. Measuring kOverall for the solid-
supported catalyst and for crushed catalyst is the most expeditious way
to determine the presence of pore diffusion. If

kOverall
Pellet
5 kOverall
Crushed

then pore diffusion is not present during product formation. If

kOverall
Pellet
, kOverall
Crushed

then product formation is pore diffusion rate limited. kOverallCrushed

is
greater than kOverall because the radius of the crushed pellet or extru-
Pellet

date is smaller than the radius of the whole pellet or extrudate, which
means reactant and product molecules spend less time traversing the
pores of the crushed sample compared to the whole sample. This
result arises because kPD is constant at controlled experimental tem-
peratures and pressures. Unfortunately, this simple experimental test
does not provide information about the extent of pore diffusion
resistance.

To determine the extent of pore diffusion resistance, consider

Eq. (2.10), which is
1 1 1 1
5 1 1
kOverall kRxn kPD AP =VP kSFD SFilm =VFilm
However, when

1 1 1
1 c
kRxn kPD AP =VP kSFD SFilm =VFilm
28 Adiabatic Fixed-bed Reactors

then the above resistance equation thus reduces to

1 1 1
5 1
kOverall kRxn kPD AP =VP
In this equation, kRxn is a constant determined from laboratory
experiments that obviated any diffusion resistance. kPD is also a con-
stant at a given operating temperature and pressure. Thus, to change
kOverall, we must change AP =VP . Catalyst manufacturers can change
AP =VP for a given catalyst; however, AP and VP are reported as distri-
butions and, while their averages for two solid-supported catalysts may
be different, their distributions will most likely overlap, thereby mak-
ing it difficult to interpret any changes in kOverall.
However, AP =VP for catalyst extrudate is

AP R2P 1
5 5
VP RP LP
2 LP

where LP is the average pore length. For a catalyst extrudate, LP can

run from the extrudate centerline to the surface, in which case it is
equivalent to the radius of the extrudate. Or, LP can run the length of
the catalyst extrudate. The former situation implies that
AP SSolid
~
VP VSolid
Thus, we can replace AP =VP with SSolid =VSolid . Note that SSolid =VSolid
is the ratio of external surface area to volume for the solid support of
the catalyst. Also, note that SSolid =VSolid ratios change with solid-
support size and shape. SSolid =VSolid is the inverse of the solids radius.
Thus, as SSolid =VSolid increases, the radius of the solid decreases.
Hence, the time required for reactant and product molecules to tra-
verse the pore network inside the solid-supported catalyst is reduced
since the pore diffusion rate constant, kPD, remains constant as the
radius changes. Result: product formation increases with increasing
SSolid =VSolid if the process is pore diffusion rate limited.
Substituting SSolid =VSolid for AP =VP in the above equation yields
1 1 1
5 1
kOverall kRxn kPD SSolid =VSolid
 Fundamentals of Fixed-Bed Reactors 29

0.1 Shapes
0.09 Trilobes
Bilobes
0.08 Modified trilobes
0.07 Quadrilobes
Cylinders
0.06
1/koverall

0.05
Sizes
0.04 1/64 in to 1/10 in
0.03
y = 45.006x 2+ 2.982x + 0.0007
0.02 R2 = 0.7635
0.01
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018
1/(S/V )

Figure 2.5 1/kOverall as a function of 1/(S/V) for Olefin isomerization by a solid-supported catalyst.

This equation suggests a linear relationship between 1=kOverall and

1=SSolid =VSolid . However, AP =VP relates to SSolid =VSolid in some
unspecified manner. We should, therefore, not expect a linear relation-
ship between 1=kOverall and 1=SSolid =VSolid . Under such a condition,
we should use the best correlation relating 1=kOverall to
1=SSolid =VSolid . Figure 2.5 shows a plot of 1=kOverall as a function of
1=SSolid =VSolid for olefin isomerization using a solid-supported cata-
lyst. Stagnant film diffusion is negligible for these experiments. A poly-
nomial provides the best fit for these data. As before, we can identify
the 1=SSolid =VSolid ratio where the resistance to product formation
moves from reaction rate limited to pore diffusion rate limited. In
Figure 2.5, at 1=SSolid =VSolid of 0.000235, 1=kOverall is 0.0014.
Subtracting 0.0007 from 0.0014 leaves 0.0007. In other words, the
resistance to reaction equals the resistance to pore diffusion. Thus, at
values of 1=SSolid =VSolid , 0.000235, the process is reaction rate lim-
ited. At all higher values, the process is pore diffusion rate limited.

2.5 IMPLICATIONS OF CONTROLLING RATE REGIMES

Pilot plant fixed-bed reactors are traditionally designed at space veloci-
ties equivalent to commercial scale fixed-bed reactors. Thus, the film
diffusion resistance of the process at the two scales is different. In gen-
eral, pilot plant fixed-bed reactors are film diffusion rate limited while
commercial-size fixed-bed reactors are either pore diffusion rate or,
30 Adiabatic Fixed-bed Reactors

more rarely, reaction rate limited. This shift from film diffusion rate
limited to, most generally, pore diffusion rate limited occurs due to the
high volumetric fluid flow through the catalyst mass in a commercial-
size fixed-bed reactor. Thus, reactant consumption or product forma-
tion is faster in the commercial-size fixed-bed reactor than in the pilot
plant fixed-bed reactor.

The same shift occurs when downsizing a process from the commer-
cial scale to the pilot plant scale. If we use a space velocity equivalent
to that of the commercial fixed-bed reactor to design a pilot plant,
then the controlling resistance to reactant consumption or product for-
mation shifts from pore diffusion or reaction rate to stagnant film dif-
fusion rate. This shift adversely impacts the results of the research
program for which the pilot plant was built. Downsized pilot plants
are built to solve process problems or to develop new catalysts. If the
process problems are related to stagnant film diffusion, then such a
pilot plant will be useful; however, if the process problems are not
related to stagnant film diffusion, then the pilot plant may produce
spurious and misleading information.

With regard to developing new catalysts, if, at the commercial scale,

the controlling resistance to reactant consumption or product forma-
tion is pore diffusion, then the median pore diameter (MPD), pore size
distribution (PSD), and SSolid =VSolid ratio for the solid support must be
optimized. However, if, at the pilot plant scale, the predominant resis-
tance to reactant consumption or product formation is stagnant film
diffusion, then changing solid support MPD and PSD will yield incon-
clusive results. Changing SSolid =VSolid will yield improved catalyst per-
formance in the pilot plant but, depending on the extent of film
diffusion resistance at the commercial scale, no improvement may be
observed when the new catalyst is adopted as the catalyst of choice.

If, at the commercial scale, the predominant resistance to reactant

consumption or product formation is reaction rate, then the chemistry
of the catalyst must be optimized. However, the results from such an
optimization program will be inconclusive if reactant consumption or
product formation in the pilot plant is controlled predominately by
stagnant film diffusion. Such a project may continue for a number of
years before the inevitable conclusion is made: that the catalyst cannot
be improved. Unfortunately, this conclusion will be drawn because the
catalysts were evaluated in the wrong flow regime.
 Fundamentals of Fixed-Bed Reactors 31

2.6 TEMPERATURE DEPENDENCE OF CONTROLLING REGIMES

All the resistances identified above are temperature dependent. At low
temperature, the reaction rate is such that no concentration difference
develops across the stagnant film or along the pores. In this case, intrinsic
surface kinetics dominates the chemical process. We can visualize the
active sites as being saturated with reactant. When the active sites are
reactant saturated, a plot of ln(kOverall) versus 1/K, where K is tempera-
ture in kelvins, is linear and its slope is 2ERxn =R.
Increasing the process temperature increases the intrinsic reaction
rate. At some temperature, a concentration difference develops along
the pores. In other words, some active sites remain saturated with reac-
tant while other active sites become reactant starved. In this case, a
plot of ln(kOverall) versus 1/K is a plot of lnkRxn 1 kPD AP =VP versus
1/K. The slope of such a plot gives the apparent activation energy for
reaction and pore diffusion. It is the arithmetic average of the energy
of activation for the reaction and the energy of activation for pore dif-
fusion. Thus, the slope is 2ERxn 1 EPD =2R. If we crush the solid-
supported catalyst, then measure kOverall as a function of temperature,
Crushed
the new lnkOverall plot will lie above the plot for lnkOverallPellet
.
11
However, their slopes will be the same.
As process temperature increases, the reaction rate becomes so high
that only those active sites on or near the catalysts surface see reac-
tant. In other words, the pores of the solid-supported catalyst are filled
with product molecules. The reactant concentration difference along
the pore disappears. The only reactant concentration difference now
occurs across the stagnant film surrounding each catalyst pellet or
extrudate. Thus, the process is stagnant film diffusion rate limited. A
plot of ln(koverall) versus 1/K is actually a plot of lnkSFD SFilm =VFilm .
The slope of such a plot is the activation energy for the diffusion of
reactant in the bulk fluid, i.e., the slope is 2EBD =R, where BD identi-
fies bulk diffusion. For gases, this EBD is 412 kJ/mole; for liquid
hydrocarbon, EBD is 1020 kJ/mole; and, for aqueous solutions, EBD
is 810 kJ/mole.11(p320)

Determining the apparent activation energy for a solid-supported

catalytic process provides another method for identifying its control-
ling resistance, which must be known in order to design equipment
and experimental programs.
32 Adiabatic Fixed-bed Reactors

2.7 DESIGN OF FIXED-BED REACTORS

Successfully scaling fixed-bed reactors requires using a kOverall for reac-
tant consumption or product formation that reflects the appropriate
process resistance at the design flow rate. Estimating such a global rate
constant is best done from Figure 2.2, which is a generalization of
Figures 2.3 and 2.4. Using such a figure to estimate kOverall at design
interstitial fluid velocity insures using the appropriate resistance.

To generate Figure 2.2 requires several fixed-bed reactor sizes.

Interstitial fluid velocities capable of minimizing stagnant film diffu-
sion resistance must be used to approach the interstitial fluid velocities
generally achieved through commercial scale fixed-bed reactors. The
interstitial fluid velocities obtainable in laboratory scale fixed-bed reac-
tors determine that portion of the relationship dominated by stagnant
film diffusion resistance. Thus, it is important to have available fixed-
bed reactors capable of operating at intermediate interstitial fluid
p
velocities to establish the relationship between 1=kOverall and 1= vz .
Once such a relationship is established, then 1=kOverall can be deter-
p
mined for any design 1= vz . 1=kOverall can then be used in a process
simulation to determine reactant consumption or product formation,
thereby yielding information as to the financial viability of the project.

2.8 SUMMARY
This chapter analyzed the performance of fixed-bed reactors and dis-
cussed the importance of identifying the resistance controlling the rate
of reactant consumption or product formation. This chapter stressed
that the controlling resistance must be identified before designing a
fixed-bed reactor or prior to planning an experimental program to sup-
port a fixed-bed, catalyzed process. This chapter also discussed the
impact of diffusion on the overall rate constant for a catalyzed process.
It also discussed the impact of temperature on the overall rate
constant.

REFERENCES
1. Carberry J. Can J Chem Eng 1958;36:207.
2. Carberry J. Chemical and catalytic reaction engineering. New York, NY: McGraw-Hill, Inc.;
1976. p. 5301.
 Fundamentals of Fixed-Bed Reactors 33

3. Smith J. Chemical engineering kinetics. 3rd ed. New York, NY: Mc-Graw Hill Book
Company; 1983. p. 477.
4. Satterfield C. Mass transfer in heterogeneous catalysis. Cambridge, MA: MIT Press; 1970.
p. 129.
5. Satterfield C. Heterogeneous catalysis in practice. New York, NY: Mc-Graw Hill Book
Company; 1980. p. 381.
6. Lee H. Heterogeneous reactor design. Boston, MA: Butterworth Publishers; 1985 [chapter 4].
7. Aris R. The mathematical theory of diffusion and reaction in permeable catalysts, vol. 1.
Oxford, UK: Clarendon Press; 1975.
8. Hougen O, Watson K. Chemical process principles: kinetics and catalysis. New York, NY:
John Wiley & Sons; 1947.
9. Hughes R. Deactivation of catalysts. London, UK: Academic Press; 1984. p. 2.
10. Granger R. Fluid mechanics. New York, NY: Dover Publications, Inc; 1995. p. 708 [First
published by Holt, Reinhart and Winston, 1985].
11. Satterfield C. Heterogeneous catalysis in practice. New York, NY: Mc-Graw Hill Book
Company; 1980. p. 321.