You are on page 1of 6

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

Contents lists available at ScienceDirect

Process Safety and Environmental Protection


journal homepage: www.elsevier.com/locate/psep

CFD modelling of an annular reactor, application to the photocatalytic degradation of acetone


Guillaume Vincent, Eric Schaer , Paul-Marie Marquaire, Orfan Zahraa
Laboratoire Ractions et Gnie des Procds, UPR CNRS 3349, Nancy Universit, 1 rue Grandville, BP 20451, 54001 Nancy Cedex, France

a b s t r a c t
This study deals with the photocatalytic degradation of acetone, which is a typical pollutant of indoor air, in an annular photoreactor. The TiO2 photocatalyst is deposited on a berglass support and irradiated by a commercial uorescent tube placed at the center of the device. Acetone conversion extents up to 90% are obtained for low initial concentrations. Neither external mass transfer nor internal diffusion limitations are observed, and the annular reactor is rst assimilated to a plug ow reactor. The Langmuir Hinshelwood model gives a good description of acetone degradation, and kinetic and adsorption parameters can be analytically deduced of the performed experiments. A Computational Fluid Dynamics description of the reactor is then proposed. The classical NavierStokes equations describe the ow in the free zone, whereas the Brinkman equation is used for the description of the ow in the porous zone. From an hydrodynamic point of view, a very good agreement between theoretical and measured Residence Time Distribution or upward velocities is obtained, testifying for the good description of the photocatalytic reactor. For a given illumination, the variations of acetone concentration inside the reactor can then be modelled. Comparison between theoretical and experimental outlet concentrations allows nally a better precision in the determination of the Langmuir Hinshelwood kinetic parameters. A relative difference of about 15% appears between the two sets of kinetic parameters (obtained assuming a plug ow or deduced from the CFD modelling). CFD simulations of photocatalytic reactor can thus address a better design of such air treatment devices. 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: CFD simulation; Photocatalytic reactor; Acetone degradation; Indoor air treatment

1.

Introduction

Photocatalytic degradation of organic compounds appears as a promising process for remediation of air polluted by VOCs. Photocatalytic processes use a semi-conductor photocatalyst, usually TiO2 , as a slurry or deposited on a support, exposed to near UV light to induce the degradation reactions. The photoinduced reaction is activated by absorption of a photon with sufcient energy, i.e. equal or higher than the band-gap energy of the catalyst (Carp et al., 2004). The adsorption leads to a charge separation due to the promotion of an electron from the valence band of the semi-conductor catalyst to the conduction band, thus generating a hole in the valence band. If the electronhole recombination is limited, the activated electron can react with an oxidant yielding a reduced product, and the generated hole can react with a reductant to produce an oxidized product.

The photocatalytic degradation kinetics is generally represented by the Langmuir Hinshelwood model. Considering that the adsorption of the reaction products and intermediates remains low, the degradation rate, proportional to the number of adsorbed reactive species, takes the form: r=k bc 1 + bc

(1)

where k is a (degradation) kinetic constant that may depend on temperature, incident light, humidity. . . and b is an adsorption constant which depends on the chemical afnity of the pollutant with the catalyst. This work deals with the study of photocatalytic degradation of acetone, which is characteristic of indoor air pollution. Acetone is naturally produced by oxidation of humic substances and it appears as a metabolic product of plants and

Corresponding author. Tel.: +33 383 17 53 04; fax: +33 383 17 53 19. E-mail address: Eric.Schaer@ensic.inpl-nancy.fr (E. Schaer). Received 30 June 2009; Received in revised form 8 March 2010; Accepted 23 August 2010 0957-5820/$ see front matter 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.psep.2010.08.004

36

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

3.
Nomenclature b C D I k r u Q V X adsorption constant (m3 kg1 ) concentration (kg m3 ) diffusivity (m2 s1 ) incident light (W m2 ) kinetic constant (kg min1 m3 ) kinetic rate (kg min1 m3 ) velocity (m s1 ) volumetric ow rate (m3 min1 ) annular reactor volume (66.4 106 m3 ) conversion extent ()

Experimental results

Under UV light and in presence of oxygen, the photocatalytic degradation of acetone can be expressed as: CH3 COCH3 + 4O2 3CO2 + 3H2 O (2)

Greek letters void fraction (0.947 here) permeability (m2 ) viscosity (Pa s) density (kg m3 ) Subscripts 0 initial

animals (Rappert and Mller, 2005). Regarding anthropogenic sources, acetone is widely used as a solvent in the paint industry, varnishes, inks and adhesives (INRS, 2003). Time weighted average values are established for acetone (1.210 g/m3 in European Union) and its photocatalytic degradation could nd potential application for reducing emissions of VOCs in the air of working space or for indoor air treatment devices. Experimental setup is rst described, experimental results are then presented, and kinetic parameters are rst determined using a plug ow model and then compared to those obtained using Computational Fluid Dynamics.

Experiments have been performed for initial acetone concentration between 0.29 and 1.73 mg L1 , incident light of 1.7 mW cm2 , relative humidities varying between 10% and 40% and feed ow rate of 300 mL min1 . Acetone conversion extents range between 20% and 90%. First results have ensured that the acetone photodegradation rate does not depend on the volumetric ow rate, indicating that the reaction (if it is not a zeroth order) is not limited by the external mass transfer. Similarly, the Weisz criterion of has been estimated at 6 105 for the photocatalytic degradation of acetone (Vincent, 2008), reecting the absence of diffusional resistance within the catalyst pellets. Residence Time Distributions have been measured in absence of reaction using a pulse of hydrogen in the feed stream and detected at the photoreactor exit by a thermal conductivity detector (TCD). The analysis of RTD revealed that the annular reactor could be assimilated to a cascade of 26 elementary continuously stirred tank reactors. Since it is generally accepted that above a number of 20 elementary reactors, a real reactor can be considered practically as a plug ow reactor, a plug ow was rst assumed in the annular reactor. For an ideal plug ow, integration of mass balance (relation (3)) allows the determination of the kinetic and adsorption parameters k and b. Indeed, the masse balance for a plug ow reactor is: r=k bc dc = 1 + bc d(V /Q ) (3)

2.

Experimental setup

and its integration leads to: V 1 = QC0 X kb ln(1 x) C0 X 1 k

The experimental setup is presented in Fig. 1. The initial air ow is split into three ways, each one controlled by a separated mass ow controller. Air is continuously bubbling through two saturators maintained under strong agitation in a thermostatic bath, one containing the volatile organic compound and the second one containing water. After owing through a mixing system, the gas stream, composed of dened quantities of air, acetone and humidity, enters the photocatalytic annular reactor. This (annular) axial symmetry conguration offers the best illumination conditions for the catalyst, and also ensures a very good contact between gas and catalyst. The catalyst is deposed on a berglass support impregnated of TiO2 Degussa P25. It is irradiated by a commercial uorescent tube placed at the center of the device, which offers dened and reproducible lighting conditions. The uorescent tube and photocatalytic support are separated by a liquid solution (a nigrosine solution) that controls the temperature inside the reactor and that also limits illumination during the photocatalytic reaction, acting as a liquid lter. The annular volume of the reactor is 66.4 cm3 . A top view sketch of the reactor is also presented in Fig. 1. A gas chromatograph equipped with a ame ionization detector is used to follow acetone concentration variations during photocatalytic oxidation.

(4)

where is the void fraction, C0 is the acetone initial concentration, X is the conversion extent, Q is the volumetric ow rate and V is the annular reactor volume. The linear variation of V/(QC0 X) against ln(1 X)/(C0 X) presented in Fig. 2 shows that the Langmuir Hinshelwood model is well adapted for the description of the photocatalytic degradation of acetone. The adsorption and kinetic constants are respectively deduced of the slope and intercept, and the obtained values are b = 8.16 L mg1 (8.16 103 m3 kg1 ) and k = 2.63 mg L1 min1 (2.63 103 kg m3 min1 ). The adsorption and kinetic constant thus obtained assuming a plug ow in the reactor give a good description of the conversion extent variations against initial concentration, as can be seen in Fig. 3a. The current reaction rate can nally be expressed as a function of acetone concentration, according to relation (1). At low concentration the kinetics is rst-order in acetone. It becomes constant (zeroth order) for larger concentrations, as can be seen in Fig. 3b. The reduction of transmitted light for different Nigrosine concentrations within the optical lter enabled studying the inuence of the incident light on the photocatalytic degradation of acetone. It appears that the adsorption constant is independent of the incident light and that the kinetic constant

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

37

Fig. 1 Experimental setup and top view sketch of the reactor.


0 -0.2 y = 0.047x-0.38 -0.4 -0.6 -0.8 -1 -10 -8 -6 -4 ln(1-X)/(C 0 X) -2 0

- V/(QC0 X)

is proportional to the incident light according to (as Wang et al., 1998) a power law: k = k0 I0 0.38 (5)

Fig. 2 Langmuir Hinshelwood parameters determination.

It is frequently reported that the values of the exponent range between 0 and 1 for most photocatalytic reactions. It has been shown (Vincent, 2008) that an exponent between 1/4 and 1/2 could be due to a recombination between OH radicals under strong illumination. Here, the measured exponent (equal to 0.38) can be attributed to mixed recombinations of electron/hole pairs and hydroxyl radicals OH during the photocatalytic oxidation of acetone. Finally, relative humidity seems to have no effect on the kinetics of reaction, for the studied range.

38

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

(a) 1
0.8 0.6

0.08

0.07 0.06 0.05

E (ts)

experimental

simulation

0.4

0.04
0.2 0 0 0.5 1 1.5
-1

Experimental Plug Flow Reactor

0.03 0.02
2

C0 (mg.L )

0.01 0

ts (s) 0 20 40 60 80 100

(b) 1.2
1 0.8 0.6 0.4 0.2 0 0

r (mg.L .min )

-1

-1

Fig. 5 Simulated and measured RTD. where u is the velocity, ( u + ( u) ) =


T

is the density and

is the viscosity. (7)

0.5

1.5
-1

C (mg.L )

The berglass support permeability has been experimentally determined by head losses measurements for different gas speeds. The estimated value is 1.1 1010 m2 .

Fig. 3 (a) Conversion extent against initial concentration for a plug ow reactor. (b) Photocatalytic degradation rate against concentration.

4.1.2.

Reaction

Hydrodynamic conditions being validated, chemical reaction within the reactor can be taken into account according to the convection and diffusion equation: D 2 C + u C = r (8)

4.
4.1. 4.1.1.

CFD simulation
Model development Hydrodynamics

4.2.

Results

The ow in the reactor is not perfectly plug (see the RTD in Fig. 5), and a better characterisation of the reactor as a better accuracy in the kinetic and adsorption parameters could be obtained with a better description of the ow. A Computational Fluid Dynamics modelling of the annular reactor has then been developed. The classical NavierStokes equations (relation (6)) describe the ow in the free zone (in the absence of support) whereas the Brinkman equation was used to describe the ow in the porous breglass support (relation (7)): + ( u) = 0 t u T + u u = [ ( u + ( u) )] t

CFD simulations of such a reactor have been performed using Comsol Multiphysics. Hydrodynamic results are rst compared with experimental ones and acetone degradation kinetic parameters are nally determined.

4.2.1.

Hydrodynamics

(6)

A full 3D model of the reactor is rst solved for a stationary laminar ow in absence of any reaction. The 3D velocity eld evolution along the reactor axis helps ensuring that the ow is relatively uniform in the photocatalytic area governed by the Brinkman equation (see Fig. 6b). The random arrangement of the berglass support allows for a fairly good distribution of gas ow and consequently a relatively homogeneous ow eld in the photoactive part of the reactor. In the annular area, integration of momentum balance for laminar ow gives the velocity distribution as follows (Bird et al., 2002): uz (r) = umax 2 (1 ln 1 r R
2

1.E-02 8.E-03 6.E-03 4.E-03 2.E-03

velocity distribution (m.s )


-1

Numerical Theoretical

2)

1 2 ln(1/)

R r

(9)

where uz (r) is the axial (upward) velocity component, r0 and R are respectively the internal and external cylinder radii, and given by: =
2.30E-02 2.35E-02 2.40E-02 2.45E-02 2.50E-02

0.E+00 2.25E-02

r0 , R

1 2 ln(1/)

(10)

radius (m)

Fig. 4 Parabolic axial velocity eld against radius in the annular zone.

Numerical axial velocity distributions in the photoactive area of the reactor can be easily deduced from CFD simulations, and as can be seen in Fig. 4, are almost superimposed

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

39

Fig. 6 Transient tracer concentration evolutions in the annular reactor at different moments.

with theoretical axial ones, testifying for the good hydrodynamic description of the reactor. Finally, a numerical RTD can be performed solving a transient tracer mass transport on top of the ow eld in the time domain. The assumption to decouple the velocity eld and the scalar transport equation is accurate because the acetone or tracer concentrations remain low. As can be seen in Fig. 5, the RTD deduced from the numerical simulation leads to a

good agreement with the experimental one, obtained with this specic photocatalytic support. The plug ow with axial dispersion observed in the photocatalytic reactor is thus perfectly described by CDF using the NavierStokes equations in the free zone and Brinkman equation in the photoactive area. The transient tracer concentration evolutions inside the annular reactor during the numerical RTD are shown in Fig. 6.

Fig. 7 Acetone concentration variations inside the annular reactor, C0 = 5 104 kg m3 .

40

Process Safety and Environmental Protection 8 9 ( 2 0 1 1 ) 3540

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.25 0.5 0.75 1

CFD, k = 2.63 mg/min/L, b = 8.16 L/mg Experimental CFD k' = 2.23 mg/min/L, b' = 6.94 L/mg

the plug ow deduced parameters induces an overestimation of the conversion extent.

5.
Conversion (X)

Conclusion

1.25

1.5

1.75

C0 (mg/L)
Fig. 8 CFD simulations with two sets of kinetic parameters.

4.2.2.

Photocatalytic degradation

The coupled simulation (hydrodynamics and chemical reaction) of acetone degradation in the annular photocatalytic reactor is nally performed using the hydrodynamic model and the previously determined kinetic parameters. It rst helps ensuring that no external diffusion effects affect the transfer, since the acetone concentration remains uniform for a given axial position. However utilisation of former kinetic parameters gives a rather poor description of the conversion and more accurate Langmuir Hinshelwood kinetic parameters of acetone degradation can nally be deduced of the comparison between theoretical and measured outlet concentrations (Vincent, 2008). The pollutant concentration evolution inside the reactor is then perfectly described, as can be seen in Fig. 7. The comparison between the kinetic parameters derived from the plug ow hypothesis and those obtained by CFD is presented in Fig. 8. A relative difference of about 15% appears between both kinetic parameters (k and b obtained assuming a plug ow and k and b deduced from CFD) but the gap remains very small. The models used in the two areas of the reactor (NavierStokes in the free zone and Brinkman in the porous zone) give a better representation of the plug ow with axial description to predict the photocatalytic conversion of acetone in the annular reactor. Results show that utilisation of

The photocatalytic degradation of acetone has been implemented in an annular reactor where the TiO2 photocatalyst is deposited on a berglass support. The reaction is not limited by the external mass transfer nor internal diffusion, and for relatively low initial concentrations conversion extents vary from 20 to 90%. The Langmuir Hinshelwood model can be applied for the description of the degradation process, and it has been shown that a CFD modelling gives a better description of the plug ow dispersion in the reactor and a better accuracy in the determination of the kinetic parameters. Indeed, application of the NavierStokes, Brinkman and convectiondiffusion equations give a pretty good match between simulation and experimental RTD and conversion variations. CFD simulations of photocatalytic reactor can thus address a better design or extrapolation of such air treatment devices and should also allow a better understanding of the physicochemical phenomena involved in photocatalytic processes.

References
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2002. Transport Phenomena, Second ed. Wiley, New York. Carp, O., Huisman, C.L., Reller, A., 2004. Photoinduced reactivity of titanium dioxide. Progress in Solid State Chemistry 32, 33177. INRS, 2003. Actone. Fiche Toxicologique 3, 8386. Rappert, S., Mller, R., 2005. Odor compounds in waste gas emissions from agricultural operations and food industries. Waste Management 25, 887. Vincent, G., 2008. Procd dlimination de la pollution de lair par traitement photocatalytique: application aux COVs. PhD. INPL, Nancy, France. Wang, K.-H., Tsai, H.-H., Hsieh, Y.-H., 1998. The kinetics of photocatalytic degradation of trichloroethylene in gas phase over TiO2 supported on glass bead. Applied Catalysis B: Environmental 17, 313320.

You might also like