DIMENSIONLESS ANALYSIS OF PHOTOCATALYTIC

REACTORS USING SUSPENDED SOLID PHOTOCATALYSTS

G. LI PUMA

Photocatalysis & Photoreaction Engineering, School of Chemical, Environmental and Mining Engineering (SChEME),
The University of Nottingham, Nottingham, UK
P
hotocatalytic oxidation (PCO) over suspended solid photocatalysts (e.g., titanium di-
oxide) has been proposed as a sustainable process for treatment and purication of
water and wastewater. However, the application of this process requires the develop-
ment of mathematical models that can be readily applied to reactor design, scale-up and
optimization.
The rigorous modelling of photocatalytic reactors requires a complex analysis of the radi-
ation eld in the photoreactor. This analysis, linked to the modelling of the uid-dynamics
and the reaction kinetics, results in integro-differential equations which almost invariably
require demanding numerical solutions. As a result these models are difcult to apply to
the analysis of photocatalytic reactors.
This paper presents a dimensionless analysis of steady-state, continuous ow, photocataly-
tic reactors using suspended solid photocatalysts that still retains the essential elements of a
rigorous approach, whilst providing simple solutions. The models developed are applicable to
either thin-lm or geometrically thick at plate and annular photoreactors, of (a) falling
lm design or (b) double-skin design, operating with three ideal ow conditions: (1) falling
lm laminar ow (FFLF); (2) plug ow (PF); and (3) slit ow (SF). The radiation eld in
the photoreactor is modelled with either a two-ux absorption-scattering model, i.e.,
scattered photons are purely back scattered, or a six-ux absorption-scattering model, i.e.,
scattered photons follow the route of the six directions of the Cartesian coordinates.
Four different dimensionless parameters appear in the models. These are the Reynolds
number, the Damkohler number, the optical thickness of the photoreactor and the scattering
albedo of the photocatalyst. The above dimensionless parameters should be maintained con-
stant during scale-up of photocatalytic reactors by dimensional analysis.
Model simulations show that at a scattering albedo higher than 0.3, radiation scattering can
signicantly affect conversions obtained at different values of optical thickness. The con-
versions with the idealized ow systems follow the sequence: FFLF . PF . SF. The
models estimate the optimum value of apparent optical thickness that maximizes conversion
in a photocatalytic reactor, which was found to be in the range from 1.8 to 3.4 depending on
ow conditions and reaction kinetics.
Keywords: photocatalysis; photocatalytic reactors; radiation scattering; thin lm; photo-
catalyst; titanium dioxide.
INTRODUCTION
Photocatalytic oxidation (PCO) over titanium dioxide
(TiO
2
) is a green sustainable process for the treatment
and purication of water and wastewater. The process
exploits the high oxidation potential of hydroxyl radicals,
which are generated at the surface of a semiconductor
photocatalyst, titanium dioxide, when this is irradiated
with UVA radiation in the presence of water and an elec-
tron acceptor. There is a large body of literature dealing
with the effectiveness of PCO for the complete destruction
of priority pollutants, reviewed by Hoffmann et al. (1995),
including aromatics, polyaromatic, halogenated hydro-
carbons, pesticides and many other classes of compounds.
However, the application of PCO for wastewater treatment
on an industrial scale is currently hindered by a lack of
simple mathematical models that can be readily applied
to reactor design and by the need for pilot-scale

Correspondence to: Dr G. Li Puma, Photocatalysis & Photoreaction Engi-

neering, School of Chemical, Environmental and Mining Engineering
(SChEME), The University of Nottingham, University Park, Nottingham
NG7 2RD, UK.
E-mail: Gianluca.Li.Puma@nottingham.ac.uk
820
02638762/05/$30.00+0.00
# 2005 Institution of Chemical Engineers
www.icheme.org/journals Trans IChemE, Part A, July 2005
doi: 10.1205/cherd.04336 Chemical Engineering Research and Design, 83(A7): 820826
demonstration studies. Current models are either too sim-
plistic or too rigorous to be useful in photocatalytic reactor
design, scale-up and optimization.
This paper presents a dimensionless analysis of steady-
state, continuous ow, photocatalytic reactors using sus-
pended solid photocatalysts that still retains the essential
elements of a rigorous approach, whilst providing simple
solutions. The models developed are applicable to either
thin-lm or geometrically thick at plate and annular
photoreactors, of (a) falling lm design or (b) double-skin
design, operating with three ideal ow conditions: (1) fall-
ing lm laminar ow (FFLF); (2) plug ow (PF); and (3)
slit ow (SF). The radiation eld in the photoreactor is
modeled with either a two-ux absorption-scattering
model, i.e., scattered photons are purely back scattered,
or a six-ux absorption-scattering model, i.e., scattered
photons follow the route of the six directions of the
Cartesian coordinates. The models were successfully vali-
dated with experimental results from the PCO of selected
pesticides in experimental reactors (Li Puma, 2003; Li
Puma et al., 2004).
THIN-FILM SLURRY PHOTOCATALYTIC REACTORS
Among the different types of scalable photocatalytic
reactors proposed in the literature, thin-lm, slurry (TFS)
photocatalytic reactors provide an excellent conguration
for efcient excitation of the semiconductor photocatalyst
TiO
2
. TFS photocatalytic reactors operate at much higher
catalyst concentrations than conventional photoreactors
because of the high opacity of TiO
2
slurry suspensions.
This property, linked to the fact that these reactors have a
very large illuminated catalytic surface area per unit
volume of reactor and minimal mass transfer limitations,
should yield an increase in the number of hydroxyl radicals
generated per unit volume of reactor, although these benet
could be somewhat moderated by the parallel increase in
the rate of hydroxyl radical recombination. The overall
effect should however be a maximization of the efciency
of photon utilization, oxidation rates and reactor through-
put. TFS photocatalytic reactors are therefore suitable for
industrial scale applications of PCO.
The most common congurations of TFS photocatalytic
reactors are that of falling lm or double-skin annular/at
reactors (Alfano et al., 2000; Li Puma and Yue, 1999)
(Figure 1). These reactor congurations can be operated
with three different ow operation modes: falling lm lami-
nar ow, plug ow and slit ow, which in turn can affect
the performance of the photoreactor. Falling lm congur-
ations have the advantage over double-skin annular/at
reactors of a very high mass transfer rate of the reducing
agent, i.e., oxygen. They also do not suffer from the lming
problem of the radiation entrance wall, which occurs in
photoreactors irradiated through a transparent window
such as in double-skin annular/at photoreactors. How-
ever, the latter reactor congurations are preferred when
the solutions to be detoxied contain high concentrations
of volatile organic carbon (VOC), which will simply be
lost from the system in falling lm reactors. Li Puma and
Yue (2003) have shown a methodology for the design of
falling lm and double-skin photoreactors.
A novel, thin-lm, slurry reactor for water treatment
and purication is the fountain photocatalytic reactor
(Li Puma and Yue, 2001ac) (Figure 2). The basic concept
of this photocatalytic reactor centres on the design of a
special nozzle that provides an unsupported thin lm in
the form of a fountain in continuous ow exposed to a
light energy source. The thin-lm is created by pumping
water through the specially designed nozzle. The liquid
lm emerges from the nozzle in the shape of a smooth
and radially expanding water fountain. The radiation source
is situated above the water fountain and can be either solar
radiation or provided by articial light sources. In the case
of solar radiation, both direct and diffuse solar radiation can
be utilized in the photoreactor. The photoreactor system has
a minimal reactor inventory as it consists of only a nozzle
and a pump and can be operated under a wide range of ow
rates, in either batch or continuous mode. It can be easily
installed in current water treatment works by the simple
distribution of these nozzles in an open-air, sunlight acti-
vated, lagoon treatment plant. In addition, the fountain
photoreactor has the ability of combining water disinfection
and water detoxication in a single process. The presence
of an unattached thin-lm water-fountain encourages a
high oxygen mass transfer rate, eliminates the need for
additional oxygenation systems, and avoids the lming
problem, sealing problem and breakage risk which are
associated with reactor congurations in which radiation
Figure 1. Geometry of TFS photoreactors. The thickness of the thin-lm is
of the order of millimetres or smaller. Figure 2. The fountain photocatalytic reactor.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
DIMENSIONLESS ANALYSIS OF PHOTOCATALYTIC REACTORS 821
enters through a transparent reactor wall. Fouling of the
reactor nozzle by the catalyst can be avoided by using an
appropriate nozzle design. The size and shape of the water
fountain can be varied by either changing the liquid ow
rate or adjusting the nozzle settings such as the radius
and aperture of the nozzle and the angle of water emerging
from the nozzle. A complete mathematical analysis of
the fountain photocatalytic reactors has been presented
(Li Puma and Yue, 2001ac) and will not be discussed
in this paper.
MODELLING OF PHOTOCATALYTIC REACTORS
The modelling of photocatalytic reactors requires a
complex analysis of the radiation eld in the photoreactor
(Cassano et al., 1995). This analysis, linked to the modelling
of the uid-dynamics and the reaction kinetics, results
in integro-differential equations which almost invariably
require demanding numerical computations. Further
advances of PCO on an industrial scale will be facilitated
by the availability of simpler mathematical models that
retain the essential elements of a rigorous model and that
can be easily used for scale-up and design.
Figure 3 shows a schematic representation of the model-
ling of a photocatalytic reactor. The development of a
reactor model requires the inclusion of a number of sub-
models. These are a radiation emission model, a radiation
absorption-scattering model, a kinetic model and a uid-
dynamic model. The central aspect of the modelling pro-
cedure is the calculation of the local volumetric rate of
photon absorption (LVRPA) at each point of the reaction
space, which requires solving the radiative transfer
equation (RTE) in the reaction space. Due to the complex
nature of radiation scattering, this results in a set of
integro-differential equations, which require demanding
numerical computational efforts. In practice, combining a
simplied radiation emission model of the light source
with a simplied radiation absorption-scattering model in
the reaction space and performing a radiation balance in
the reaction space can simplify the RTE.
The above scheme assumes that the useful photons of a
given photocatalytic reactor, i.e., those photons with energy
higher than the band-gap of the semiconductor photo-
catalyst, are absorbed by the solid photocatalytic particles
only. This assumption removes the interdependence of the
progress of the reaction and the attenuation of the radiation
because the absorbing species do not undergo changes in
concentration. Consequently, the incident radiation ux
becomes a function of the reactor position only and can
be obtained independently of the information provided by
the material balance equation.
Once the LVRPA has been calculated, it is normally
substituted into the kinetic equation and into the material
balance equation which, when solved with suitable bound-
ary conditions, yields the concentration of a generic
substrate j at the reactor outlet.
Three approaches have been proposed in the literature
for the calculation of the LVRPA: (1) the rigorous
approach which involves the mathematical solution of
the RTE, although its integro-differential nature makes this
approach considerably complex (Cassano et al., 1995); (2)
the numerical approach which involves the Monte Carlo
simulation of the radiation eld in the photoreactor, a
simple but also a computationally demanding procedure
(Pasquali et al., 1996); and (3) the simplied approach
which models the radiation eld in the photoreactor using
two-ux (Raupp et al., 1997; Brucato and Rizzuti,
1997), or six-ux (Li Puma et al., 2004) radiation
absorption-scattering models. The two and six-ux
models yield a sensible representation of the LVRPA in
the reaction space and allow a considerable simplication
of the mathematical model. Furthermore, computational
time in a standard PC would be signicantly reduced
from, for example, the order of several hours or days
using the Monte Carlo method, to a few minutes using
the two and six-ux models.
The scattering properties of the photocatalyst, and the
geometrical conguration of the photoreactor, determine
in large extent the degree of complexity of a mathematical
model. In this paper, three different levels of complexity of
slurry photocatalytic reactors models are discussed as
shown in Table 1.
DIMENSIONLESS PARAMETERS
In the models in Table 1 and references therein, four
different dimensionless parameters appear. These are, the
Reynolds number, N
Re
, which suggests the type of uid-
dynamic model (laminar or turbulent ow) to be included
in the model, the Damkohler number, N
Da
which has the
meaning of the ratio of the overall reaction rate, calculated
at the inlet concentration and at maximum photon ux, to
the maximum input mass ow rate of the reactant.
N
Da
1 indicates that the disappearance of the substrate
j is much slower than the convective transport through
the reactor. The optical thickness of the photoreactor t,
which is a measure of the degree of opacity of the photo-
reactor, and the scattering albedo of the photocatalyst, v,
Figure 3. Schematic representation of the modelling of a photocatalytic reactor.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
822 LI PUMA
which relates to the optical properties of the photocatalyst.
The above dimensionless parameters should be maintained
constant during scale-up of photocatalytic reactors by
dimensional analysis.
Level 1: TFS Photocatalytic Reactors with v < 0.3
An important parameter in the modelling of a photocata-
lytic reactor is the scattering albedo of the photocatalyst, v.
This is a dimensionless parameter equal to the ratio of the
scattering coefcient, divided by the extinction coefcient
of the photocatalyst:
v
s
s k
(1)
where s and k are the specic mass scattering and absorp-
tion coefcients respectively, averaged over the useful
spectrum of the incident radiation:
s
_
l
max
l
min
s
l
I
l
dl
_
l
max
l
min
I
l
dl
(2)
k
_
l
max
l
min
k
l
I
l
dl
_
l
max
l
min
I
l
dl
(3)
where l
min
and l
max
are respectively the minimum and
maximum wavelengths of the incident radiation that can
be absorbed by the photocatalyst. Note that the scattering
albedo should be evaluated using the absorption and scat-
tering coefcients determined experimentally under the
prevailing conditions of the slurry suspension (i.e., with
the observed degree of agglomeration of TiO
2
in the
suspension).
Model simulations have shown that when the value of
the scattering albedo of the photocatalyst is 0.3 or smaller,
in practice the effect of radiation scattering can be safely
neglected in the modelling procedure (Li Puma, 2003).
Thus mathematically this case can be re-conducted to
level 2 of model complexity, with v 0. Under the
above circumstance, the dimensionless LVRPA takes the
form:
LVRPA

0;z
exp (tj

) (4)
which is similar to the Beer Lambert law. As a result, a
simple dimensionless analysis of the reactor can be per-
formed as presented in Li Puma and Yue (2003). In
equation (4), t is the optical thickness of the photoreactor,
a fundamental dimensionless parameter in the design of
photocatalytic reactors which is dened as follows:
t (s k)C
cat
d (5)
where d is the thickness of the reaction space, C
cat
is the
photocatalyst concentration, and
I

0;z

I
j

0;z

(I
max
j

0
)
v0
(6)
is the dimensionless intensity of the incident radiation
at the entrance window of the photoreactor, where
(I
max
j

0
)
v0
is the maximum value of the radiation intensity
at the entrance window of the photoreactor calculated in
the absence of radiation scattering. In a symmetrical photo-
reactor system, (I
max
j

0
)
v0
corresponds to the value of the
incident ux in the forward direction calculated at the
front window of the photoreactor at the middle length
(z

0.5), where z

is the dimensionless axial direction

of the photoreactor.
Although this model is easy to use, its applicability in the
modelling of photocatalytic reactors is rather limited since
commercial photocatalysts (i.e., TiO
2
powders) have scat-
tering albedos higher than 0.5. This implies that the radi-
ation scattering phenomena should be included in the
modelling of photocatalytic reactors, which suggests that
level 2, or level 3 of complexity of slurry photocatalytic
reactors models must be used. Nevertheless, the above
model can be useful for an optimistic analysis of the
performance of photocatalytic reactors. Since radiation
scattering always reduces the number of photons absorbed
within the reaction space of a photocatalytic reactor, it can
be expected that the performance of a photocatalytic
reactor in the presence of scattering will be worse com-
pared to the case in which radiation scattering has been
neglected (see also Figure 5 which is discussed later).
Level 2: TFS Photocatalytic Reactors, any Value of v
In this model the effect of radiation scattering by sus-
pended solids, i.e., the photocatalyst, is taken into account.
A suitable radiation absorption-scattering model for TFS
photocatalytic reactors is the two-ux radiation model
(Brucato and Rizzuti, 1997). This is a one-dimensional
radiation model in which the incident radiation at a point
(j

, z

) in the liquid lm is taken to be equal to the sum

of the radiation that travels in the forward direction and
the fraction that travels in the backward direction due to
back-scattering (Figure 4). The travelling direction of
these photons is taken to be orthogonal to the main liquid
ow direction.
A dimensionless analysis of steady-state, continuous
ow, TFS photoreactor congurations for falling lm and
Table 1. Levels of complexity of slurry photocatalytic reactors models. v
is the scattering albedo of the photocatalyst.
Mathematical
complexity
Level 1: TFS photocatalytic reactors with v , 0.3
(Li Puma and Yue, 2003)
Lower
Level 2: TFS photocatalytic reactors, any value of v
(Li Puma, 2003)
Intermediate
Level 3: Geometrically thick slurry photocatalytic
reactors, any value of v (Li Puma et al., 2004)
Higher
Figure 4. The two-ux model.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
DIMENSIONLESS ANALYSIS OF PHOTOCATALYTIC REACTORS 823
double-skin operating with three ideal ow conditions: (1)
falling lm laminar ow (FFLF); (2) plug ow (PF); and
(3) slit ow (SF) has been shown in Li Puma (2003)
which is applicable to optically thick reactors (t
app
1).
The dimensionless LVRPA at a point (j

, z

) in the reac-
tion space calculated with the two-ux model is:
LVRPA

0;z
exp (t
app
j

) (7)
where
LVRPA
max
kC
cat
f (v; t)(I
max
j

0
)
v0
(8)
f (v; t) 1
1
v
1
1 c
1 c

1 v
2
p
_ _
(9)
c
1

1 v
2
p
1

1 v
2
p exp 2t

1 v
2
p
_ _
(10)
t
app
t

1 v
2
p
(11)
f (v, t) is a dimensionless function introduced to account
for radiation scattering using the two-ux model. This func-
tion depends on the scattering albedo v and on the optical
thickness t. The expression for f (v,t) for the case of pure
back-scattering can be derived from the study of Brucato
and Rizzuti (1997) which provides an approximate tool to
model the radiation eld. For the case of a purely absorbing
medium (v 0), f (v, t) reduces to 1 and the expression for
the radiation intensity in the reaction space reduces to the
Beer Lamberts law. Conversely, for a purely scattering
medium (v 1) with s !1, f (v, t) 2 and all incident
radiation is back-scattered at the front windowof the reactor.
The apparent optical thickness, t
app
, is the apparent (or
observed) optical thickness that would be calculated by
measuring the forward radiation at the front and at the back
wall of the photoreactor, and takes the meaning of observed
opacity of the reactor.
Figure 5 shows an example of the model simulations for
the conversion of a substrate j as a function of the apparent
optical thickness at different values of the scattering albedo,
respectively for FFLF, PF and SF ideal ow systems.
This shows that there exists an optimal value of the
apparent optical thickness, which maximizes conversion
in a photocatalytic reactor. Further simulations results
shows that this optimum value is in the range from 1.8 to
3.4 depending on ow conditions and reaction kinetics
(Li Puma, 2003). This gives a heuristic for the design of
photocatalytic reactors.
Figure 5 also shows that radiation scattering signicantly
affects the conversions obtained at different values of the
apparent optical thickness. As the scattering albedo v is
increased the conversions decrease signicantly because
of losses of photons from the reacting systems due to radi-
ation scattering.
Finally, the model simulations show that independently
from the values of the scattering albedo and apparent opti-
cal thicknesses, the conversions obtained with the three
different idealized ow systems follow the sequence:
FFLF . PF . SF. The reasons for this have been
explained in full elsewhere (Li Puma, 2003; Li Puma and
Yue, 2003) and can be attributed to the degree of corre-
spondence of radiation eld and uid residence time. For
example, in the FFLF system there is a very good degree
of correspondence of radiation eld and uid residence
time since the uid elements with the lowest residence
times near the front wall (j 0) receive maximum inten-
sity of radiation, and the uid elements with the longer resi-
dence times near the back wall (j 1) receive the lowest
intensity of radiation.
Level 3: Geometrically Thick Slurry Photocatalytic
Reactors, any Value of v
Geometrically thick photoreactors are those reactors in
which the thickness of the reaction space cannot be neglec-
ted in comparison with the other dimensions of the reactor.
An example is the annular photoreactor used in many
laboratories for kinetic studies.
To model these types of reactors we have introduced a
novel six-ux absorption-scattering model (SFM), which
Figure 6. The six principal directions of photons scattering in the six-ux
model and the scattering probabilities.
Figure 5. Model simulations for the conversion of a substrate j as a function
of the optical thickness and scattering albedo for different idealized ow
conditions. Model parameters are as in Li Puma (2003) table 4, with kinetic
parameters m 1, n 1, k
T
2 10
25
kg
(12n)
s
21
W
2m
m
(3m3n23)
.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
824 LI PUMA
is a three-dimensional extension of the two-ux model,
used to model the TFS photoreactors. The SFM allows an
integral analytical estimation of the local volumetric rate
of photon absorption (LVRPA) at each point of the reactor.
The SFM has shown close agreement with the Monte
Carlo simulation of the radiation eld in a planar photo-
reactor. This model takes into account both absorption
and scattering of photons. The main assumption of this
model is that photons are either absorbed or scattered
upon colliding with a particle, and scattering follows the
route of one of the six directions of the Cartesian coordi-
nates (Figure 6).
A complete mathematical analysis of the SFM is
reported in Li Puma et al. (2004). The dimensionless
LVRPA at a point (r

) in the reaction space of an innitely

long annular photocatalytic reactor calculated with the
SFM in cylindrical coordinate is
LVRPA

0;z

h
h(1 h)j

v
corr
1

1 v
2
corr
_
_ _
exp(t
app
j

)
g v
corr
1

1 v
2
corr
_
_ _
exp(t
app
j

)
_
_
_
_
_
_

_
v
corr
1

1 v
2
corr
_
_ _
g v
corr
1

1 v
2
corr
_
_ _
_
_
_
_
_
_

_
(12)
where
j

r hR
RhR
(13)
a 1 vp
f

4v
2
p
2
s
(1 vp
f
vp
b
2vp
s
)
(14)
b vp
b

4v
2
p
2
s
(1 vp
f
vp
b
2vp
s
)
(15)
v
corr

b
a
(16)
g
1

1 v
2
corr
_
1

1 v
2
corr
_ exp
2R(1 h)
l
vcorr
_ _
(17)
LVRPA
max

(I
max
j

0
)
v0
l
vcorr
v
corr
(1g)
v
corr
1

1 v
2
corr
_
_ _ _
g v
corr
1

1 v
2
corr
_
_ __
(18)
t
app
at

1 v
2
p
(19)
The model parameters p
b
, p
f
and p
s
of the SFM are the
probability of backward, forward and side scattering.
These can be estimated by tting the SFM to the exact solu-
tion of the radiative transfer equation by a Monte Carlo
approach. For the case of a diffusively reecting large
sphere phase function (Cassano et al., 1995) given by:
F(w)
8
3p
( sinww cos w) (20)
p
b
, p
f
and p
s
were found to be equal to 0.71, 0.11 and 0.045,
respectively (Brucato et al., submitted).
The six-ux model is clearly mathematically more com-
plex than the two-ux model. Nevertheless such a degree of
complexity is necessary for the correct analysis of annular
photocatalytic reactors. An experimental validation of the
SFM has been presented in Li Puma et al. (2004).
Figure 7 shows model simulation results for the radial
proles of the LVRPA at the middle length of an annular
photocatalytic reactor at different optical thicknesses calcu-
lated using the two- and the six-ux models. These show
that as the value of the optical thickness of the photoreactor
increases, the radial prole of the LVRPA becomes steeper
near the radiation entrance wall and less intense near the
back of the thin-lm which is consistent with Monte
Carlo simulations of the radiation eld of the photoreactor
(Pasquali et al., 1996).
Figure 7 also shows that there are notable differences
between the radial proles of the LVRPA calculated with
the two-ux model and the six-ux model. The two-ux
model underestimates the LVRPA and the degree of under-
estimation increases as the optical thickness increases. This
is due to the larger degree of photon absorptions which the
SFM predict since photons can reach a certain point from
lateral and axial scattering as well as from forward and
back scattering. Further simulations have conrmed that
the Monte Carlo simulation of the radiation eld in a
planar photoreactor ts the SFM rather than the two-ux
model (Brucato et al., submitted for publication).
CONCLUSIONS
This paper has presented an analysis of steady-state con-
tinuous ow slurry photocatalytic reactors (SPR) for water
purication by TiO
2
photocatalysis. It has shown mathe-
matical models in dimensionless form that can be easily
used for scale-up and design of photoreactors of different
geometries irradiated by UV lamps or by solar radiation.
The effect on reactor performance of radiation scattering
by the solid photocatalyst has been emphasized throughout
the paper. The models developed are applicable to either
thin-lm or geometrically thick at plate and annular
photoreactors, of (a) falling lm design or (b) double-skin
Figure 7. Radial proles of the LVRPA at the middle length of an annular
photocatalytic reactor at different optical thicknesses calculated with the
two- and the six-ux models. Experimental conditions and reactor dimen-
sions are as those reported in Li Puma et al. (2004).
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
DIMENSIONLESS ANALYSIS OF PHOTOCATALYTIC REACTORS 825
design, operating with three ideal ow conditions: (1) fall-
ing lm laminar ow (FFLF); (2) plug ow (PF); and
(3) slit ow (SF). The radiation eld in the photoreactor
is modelled with either a two-ux absorption-scattering
model, i.e., scattered photons are purely back scattered,
or a three-ux absorption-scattering model, i.e., scattered
photons follow the route of the six directions of the
Cartesian coordinates. The models are expressed in dimen-
sionless form and scale-up of SPR photocatalytic reactors
can be carried out by dimensional analysis. In addition,
the model parameters can be estimated easily from real sys-
tems and model solutions can be obtained with modest
computational efforts.
Current models of photocatalytic reactors available in the
literature are either too rigorous or too simplistic to be used
effectively for the design and scale-up of photocatalytic
reactors. The present models retain the essential elements
of a rigorous approach, whilst providing simple solutions,
which will aid in the development of photocatalytic oxida-
tion technology for water treatment and purication.
NOMENCLATURE
C
cat
photocatalyst concentration, kg m
23
H length of the reactor, m
I radiation intensity (or radiative ux), W m
22
I
l
radiation intensity divided wavelength of radiation, W m
23
I
max
j0
maximum value of radiation intensity at surface j 0,
W m
22
I

;z
dimensionless radiation intensity ( I
j;z
=I
max
j0
)
LVRPA

dimensionless LVRPA,
[ (kC
cat
I
j;z
)=(kC
cat
I
max
j0
) LVRPA=LVRPA
max
]
N
Da
Damkohler number, dimensionless
N
Re
Reynolds number, dimensionless
p
b
probability of scattering in the backward direction,
dimensionless
p
f
probability of scattering in the forward direction,
dimensionless
p
s
probability of scattering in the side direction,
dimensionless
r radial coordinate, m
r
l
lamp radius, m
R external radius of annulus, m
z axial coordinate, m
z

dimensionless axial coordinate (z/H)

Greek symbols
g parameter dened by equation (17), dimensionless
d thickness of reaction space, m
h ratio of internal radius to external radius of annulus,
dimensionless
k specic mass absorption coefcient averaged over the
spectrum of the incident radiation, m
2
kg
21
l radiation wavelength, m
j reaction space transversal coordinate, m
j

dimensionless reaction space transversal coordinate,

(j/d), dimensionless
s specic mass scattering coefcient averaged over the
spectrum of the incident radiation, m
2
kg
21
c parameter dened by equation (20), dimensionless
t optical thickness, dimensionless
t
app
apparent optical thickness, equation (11) for TFM or
equation (19) for SFM, dimensionless
w scattering angle
F scattering phase function, dimensionless
v scattering albedo (s/(s k)), dimensionless
v
corr
corrected scattering albedo, equation (16), dimensionless
Subscripts
l wavelength
j

0 position at inner wall of annulus

z, z

direction along the axial coordinate

j, j

direction along the lm transversal coordinate

v 0 calculated in the absence of radiation scattering
max maximum value
Superscripts

dimensionless variable
max maximum value
REFERENCES
Alfano, O.M., Bahnemann, D., Cassano, A.E., Dillert, R. and Goslich, R.,
2000, Photocatalysis in water environments using articial and solar
light, Catal Today, 58: 199230.
Brucato, A. and Rizzuti, L., 1997, Simplied modeling of radiant elds in
heterogeneous photoreactors. 2. Limiting two ux model for the case
of reectance greater than zero, Ind Eng Chem Res, 36: 47484755.
Brucato, A., Cassano, A.E., Grisa, F., Montante, G. and Rizzuti, L. A six-
ux model for the estimation of radiant elds in at heterogeneous
photoreactors, AIChE J, (submitted for publication).
Cassano, A.E., Martin, C.A., Brandi, R.J. and Alfano, O.M., 1995, Photo-
reactor analysis and design: fundamentals and applications, Ind Eng
Chem Res, 34: 21552201.
Hoffmann, M.R., Martin, S.T., Choi, W. and Bahnemann, D.W., 1995,
Environmental applications of semiconductor photocatalysis, Chem
Rev, 95: 6996.
Li Puma, G., 2003, Modeling of thin-lm slurry photocatalytic reactors
affected by radiation scattering, Env Sci Technol, 37: 57835791.
Li Puma, G. and Yue, P.L., 1999, Comparison of the effectiveness of
photon-based oxidation processes in a pilot falling lm photoreactor,
Env Sci Technol, 33: 32103216.
Li Puma, G. and Yue, P.L., 2001a, A novel fountain photocatalytic reactor
for water treatment and purication: modeling and design, Ind Eng
Chem Res, 40: 51625169.
Li Puma, G. and Yue, P.L., 2001b, A novel fountain photocatalytic reactor:
model development and experimental validation, Chem Eng Sci, 56:
27332744.
Li Puma, G. and Yue, P.L., 2001c, The modeling of a fountain photo-
catalytic reactor with a parabolic prole, Chem Eng Sci, 56: 721726.
Li Puma, G. and Yue, P.L., 2003, Modelling and design of thin-lm slurry
photocatalytic reactors for water purication, Chem Eng Sci, 58: 2269
2281.
Li Puma, G., Khor, J.N. and Brucato, A., 2004, Modeling of an annular
photocatalytic reactor for water purication: oxidation of pesticides,
Env Sci Technol, 38: 37373745.
Pasquali, M., Santarelli, F., Porter, J.F. and Yue, P.L., 1996, Radiative
transfer in photocatalytic systems, AIChE J, 42: 532537.
Raupp, G.B., Nico, J.A., Annangi, S., Changrani, R. and Annapragada, R.,
1997, Two-ux radiation-led model for an annular packed-bed photo-
catalytic oxidation reactor, AIChE J, 43: 792801.
ACKNOWLEDGEMENTS
The author is grateful to NATO (Grant SfP-977986) for nancial
support.
This paper was presented at the 7th World Congress of Chemical
Engineering held in Glasgow, UK, 1014 July 2005. The manuscript
was received 15 December 2004 and accepted for publication after
revision 18 April 2005.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A7): 820826
826 LI PUMA