SOLARSAFEWATER SYMPOSIUM, P. Iguazú, Argentina
October 17-19, 2005
MODELING OF A SOLAR REACTOR FOR
WATER PURIFICATION, EMPLOYING
THE PHOTO-FENTON REACTION
Germán H. Rossetti, Enrique D. Albizzati,
and Orlando M. Alfano
Universidad Nacional del Litoral - CONICET
Güemes 3450, 3000 Santa Fe, Argentina
E-mail: alfano@intec.unl.edu.ar
OUTLINE
N
N
N
N
N
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
INTRODUCTION
N
N
¾
¾
The Fenton reaction is a chemical system involving
hydrogen peroxide and ferrous salts that generates
highly reactive hydroxyl radicals.
The oxidation ability of the Fenton mixture can be
greatly enhanced using UV (or UV/Vis) radiation: the
photo-Fenton Reaction.
In this work, the degradation of formic acid (a model
pollutant) in aqueous solution using the Fenton and
photo-Fenton systems is presented.
The reaction was conducted in a flat-plate solar reactor
placed inside the loop of a batch recycling system.
INTRODUCTION: PREVIOUS WORK
Modeling and Experimental
Verification of a Flat-Plate Solar
Photoreactor, Ind. Eng. Chem.
of Formic Acid in a
Res., Decomposition
37, 3592 (1998).
Water Solution Employing the
Photo-Fenton Reaction, Ind. Eng.
Chem. Res., 41, 1436 (2002).
Modeling of a Flat-Plate Solar
Reactor. Degradation of Formic Acid
by the Photo-Fenton Reaction, Solar
Energy, 77, 461(2004).
Temperature Effects on the Photo-Fenton
Degradation of Formic Acid, ENPROMER 2005,
Río de Janeiro, Brasil; III EPOA, Campinas, Brasil.
FLOW SHEET OF THE
EXPERIMENTAL DEVICE
9
4
2
10
6
3
7
1
8
5
Keys: (1) storage tank, (2) stirrer, (3) thermometer, (4) liquid
sampling, (5) pump, (6) valve, (7) solar radiation, (8) flat-plate
reactor, (9) heat exchanger, and (10) thermostatic bath.
PICTURE OF THE EXPERIMENTAL DEVICE
N
N
N
Well-stirred batch
recycling photoreactor
Flat-plate solar reactor
Broadband UV Radiometer
CUV3 of Kipp & Zonen
OUTLINE
N
)
N
N
N
N
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
MASS BALANCES
dC F
=
dt
VR
VT
R F (x , t )
VR
photo-Fenton
dC P
dt
=
VR
VT
R P (x , t )
VR
photo-Fenton
Initial conditions:
t=0
t=0
C F = C 0F
C P = C 0P
(F: formic acid; P: Hydrogen Peroxide)
⎛ VTk ⎞ t
⎟ R F (t )
+⎜
⎜V ⎟
⎝ T ⎠
Fenton
N
VR/VT for photo-Fenton
N
VTk/VT for Fenton
N
The average value must
be retained in order to
account for spatial
variations of the photoFenton reaction rate
N
Spatial variations of the
Local Volumetric Rate
of Photon Absorption
(LVRPA)
⎛ VTk ⎞ t
⎟ R P (t )
+⎜
⎜V ⎟
⎝ T ⎠
Fenton
OUTLINE
N
N
)N
N
N
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
REACTION SCHEME(*)
Initiation
Propagation
Termination
Decomposition
hν
+ H2O → Fe2+ + HO• + H+
Fe3+ + H2O2 → Fe2+ + HO2• + H+
Fe2+ + H2O2 → Fe3+ + HO• + HOH2O2 + HO• → HO2• + H2O
H2O2 + HO2• → HO• + H2O + O2
2 HO• → H2O2
2 HO2• → H2O2 + O2
HO2• + HO• → H2O + O2
Fe3+ + HO2• → Fe2+ + H+ + O2
Fe2+ + HO2• + H+ → Fe3+ + H2O2
HCOOH + HO• → CO2•- + H2O + H+
CO2•- + O2 + H+ → CO2 + HO2•
Fe3+
(*) Proposed by Pignatello (1992), De Laat and Gallard (1999)
φ
k1
k2
k3
k4
k5
k6
k7
k8
k9
k10
k11
ASSUMPTIONS FOR THE KINETIC MODEL
The following assumptions have been considered:
N
N
N
N
the steady state approximation (SSA) may be
applied for highly reactive radicals, such as OH•
and HO2•,
radical-radical termination reactions are neglected
as compared with the propagation reactions,
the ferrous ion concentration remains constant
during the reaction time,
the oxygen concentration is always in excess.
KINETIC MODEL
1/ 2
⎛ Φ ∑ e (x, t ) ⎞ ⎛ Φ ∑ e (x, t ) ⎞
⎟
⎟ + ⎜1 + λ
λ
R F ( x , t ) = − ⎜⎜
⎜ 1 + K 3 (C P C F ) ⎟⎟ ⎜⎜
K 4 C Fe3+ C P ⎟⎟
⎝
⎠ ⎝
⎠
a
λ
a
λ
R Ft ( t )
Φ : wavelength-averaged primary quantum yield
: spectral LVRPA
K i : kinetic parameters (i = 1 to 4)
e aλ ( x , t )
When Σλ ea(x,t) = 0, the pollutant reaction rate is not null.
A thermal reaction rate can be identified (Fenton reaction).
This term may be represented by the expression:
R ( t ) = − K1
t
F
1 + K 2 (C P C Fe3+ )
1 + K 3 (C P C F )
C Fe3+ C P
OUTLINE
N
N
N
)N
N
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
RADIATION FIELD MODELING
METALLIC qD
FRAME
θ*i
GLASS
PLATE
AIR
θ 'cr
qS
θ'r
x=0
INLET
x
x=L
N
N
N
θr
θcr
θ cr'
OUTLET
REACTION SPACE
ρBqB
Schematic representation of the flat-plate solar reactor
At the top, a window made of glass was located
The surface of radiation entrance receives direct solar
radiation (qD) and diffuse solar radiation (qS)
RADIATION FIELD MODELING
Radiative Transfer Equation:
μ
∂I λ ( x , μ, φ)
∂x
+ κ λ I λ ( x , μ, φ) = 0 μ = cos θ
Variation of the Radiation
Intensity along the ray path
Î
Î
Radiation
absorption
B.C. at x = 0: (i) reflection and refraction at the interfaces
and (ii) radiation absorption inside the glass window
B.C. at x = L: radiation intensity reaching the reactor bottom
is reflected back to the solution in a diffuse manner
EVALUATION OF THE LVRPA
Reaction
Space
I(θ,φ)6
I(θ,φ)1
Once the radiation intensity Iλ(x,μ,φ) is
obtained, one can compute the LVRPA:
Radiation may be arriving at one
point (P) inside the reaction space
from all directions in space
An integration over all the arriving
rays (θ,φ) is required:
N
P
I(θ,φ)5
I(θ,φ)2
I(θ,φ)4 I(θ,φ)3
N
2π
λ 0
e ( x ) = κ ∫ dφ ∫ I λ ( x, μ, φ) dμ
a
λ
1
-1
Integrating the previous equation, LVRPA is obtained:
FINAL EXPRESSION OF THE LVRPA
[
]
∗
⎧⎪
1
−
ρ
(
μ
ap
i ) [1 − ρ p w (μ′r )]τ λ (μ′r )
a
exp( − κ λ x / μ r ) +
e λ ( x ) = κ λ ⎨q D ,λ
2
1 − τ λ (μ′r ) ρ a p (μ′r ) ρ p w (μ′r )
⎪⎩
−
−
−
−
Direct solar radiation
2q S,λ
1
n 2w
∫
n a2
μ cr
[1 − ρa p (μ∗ )][1 − ρ p w (μ′)]τλ (μ′) exp(− κ
−
−
1 − τ λ2 (μ′) ρ a p (μ′) ρ p w (μ′)
−
λ x / μ )dμ +
−
Diffuse solar radiation
1
⎡
2ρ B q B,λ ⎢ ρ w p (μ ) exp[ − κ λ ( L + x ) / μ ]dμ +
⎣ 0
∫
−
⎫
⎤
exp[ − κ λ ( L − x ) / μ ]dμ ⎥ ⎬
0
⎦⎭
∫
1
Radiation flux at the reactor bottom
OUTLINE
N
N
N
N
)
N
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
SOLAR RADIATION INCIDENT
AT THE REACTOR WINDOW
Sun
θZ
Extraterrestrial
Radiation
Atmosphere
Direct radiation
Diffuse radiation
Absorption and
Scattering
Global radiation
Ozone
Oxygen
Nitrogen
Carbon Dioxide
Water Vapor
Aerosols
SOLAR RADIATION INCIDENT
AT THE REACTOR WINDOW
Î Global
radiation on a horizontal surface at ground level
for wavelength λ (Bird and Riordan, 1986):
q G ,λ = q D ,λ cos θZ + q S,λ (θZ = zenith angle)
N
Direct radiation on a surface normal to the sun direction:
q D ,λ = H 0,λ D Tr ,λ Ta ,λ Tw ,λ To ,λ Tu ,λ
N
Diffuse radiation on a horizontal surface at ground level:
q S, λ = q r , λ + q a , λ + q g , λ
where: Rayleigh scattering (qr,λ), aerosol scattering (qa,λ), multiple
reflection of radiation between the ground and the air (qg,λ)
GLOBAL AND DIFFUSE
UV SOLAR RADIATION
N
40
q i (W/m 2)
Rosario
Argentina
qG
Measurements and
model predictions:
¾
¾
qS
20
N
Model predictions:
¾
¾
N
0
10
30
50
θ z (º)
N
N
70
90
horizontal surface
clear sky days
Global (
Diffuse (
)
)
Measurements:
¾
¾
Global (◊)
Diffuse (†)
Maximum UV solar radiation: qG,max ≅ 45 W/m2
At θZ > 45°
Diffuse radiation > Direct radiation
RATIO OF UV TO TOTAL SOLAR
RADIATION (R)
Measurements (◊)
and predictions ( ):
0.06
0.05
Ö horizontal surface
Ö clear sky days
R(−)
0.04
0.03
0.02
Rosario
ROSARIO
Año 1995
Argentina
0.01
0.00
0
10
20
30
40
50
θz(º)
N
N
60
70
80
90
θZ(°)
R%
12
20
40
60
80
5.2
5.1
4.8
4.2
3.9
UV solar radiation: 4 to 5% of the total solar radiation
R decreases when the zenith angle is increased
N
α Fe (OH )2+
qG
qD
κP
q i (W m−2 nm−1 )
κ P (cm −1), α Fe(OH )2+ × 10−2 (m2 mol −1)
SPECTRAL DATA
¾ cloudless
sky conditions
¾ solar zenith angle = 10°
qS
λ (nm)
Global (qG), direct (qD)
and diffuse (qS) solar
radiation (Bird and
Riordan, 1986) for:
N
Molar absorptivity of the
iron complex: αFe(OH)2+
N
Absorption coefficient
of the glass plate: κP
NUMERICAL SOLUTION:
COMPUTATIONAL STEPS
Evaluation of the direct and
diffuse solar radiation incident
at the reactor glass
window of the LVRPA
Computation
as a function of position
Evaluation of the formic
acid and hydrogen peroxide
reaction
rates
Calculation of the formic
acid and
hydrogen
peroxide concentrations as a function of time
System of two nonlinear, first order,
ordinary differential equations
OUTLINE
N
N
N
N
N
)
N
N
N
Introduction
Mass Balances
Kinetic Model
Radiation Field
Model Parameters and Numerical Solution
Predicted and Experimental Results
Effects of the Reaction Temperature
Final Remarks
e a × 109 (einscm −3 s−1 )
PREDICTIONS OF THE LVRPA
LVRPA as a function
of the x-coordinate for:
C Fe( OH ) 2+ = 1 mM
θ z = 10º
N
30º
N
60º
0.0
0.5
1.0
1.5
2.0
2.5
Three different zenith
angles: 10°, 30°, 60°
Constant absorbing
species concentration:
CFe(OH)2+ = 1 mM
x (cm)
Î
As expected, the radiation field along the x-coordinate
is highly non-uniform: ea(x = 0.5 L) ≅ 0.2 ea(x = 0)
e a × 109(einscm −3 s−1 )
PREDICTIONS OF THE LVRPA
θ z = 10º
30
C Fe ( OH ) 2+ = 2 mM
20
N
three different ferric
ion concentrations:
CFe(OH)2+ = 0.5, 1, 2 mM
N
a constant solar zenith
angle: θz = 10°
1mM
10
0.5mM
0
0.0
0.5
1.0
1.5
x (cm)
Î
ea as a function of the
x-coordinate for:
2.0
2.5
When the optical density is increased the shape of the
LVRPA curve becomes steeper
PREDICTED AND EXPERIMENTAL
RESULTS (T = 25 °C)
N
1.00
C i C 0i
0.75
N
0.50
Model predictions
and experimental data
as a function of time
Formic acid relative
concentration:
¾ Fenton
(
)
¾ photo-Fenton (
0.25
CP/CF = 3.3
θz = 12.8°
0.00
0
20
N
40
Time (min)
60
)
H2O2 relative
concentration:
¾ Fenton
(
)
¾ photo-Fenton (
)
PREDICTED AND EXPERIMENTAL
RESULTS (25 °C)
1.00
N
Ci Ci0
0.75
N
0.50
0.25
N
CP/CF = 8.4
θz = 12.8°
0.00
0
20
40
Time (min)
60
N
A similar representation is
shown for a higher CP/CF
Conversion for the photoFenton reaction is always
higher than that obtained
with the Fenton reaction
Model and experimental
results show good
agreement
The maximum error is 9%
COMPARISON BETWEEN FENTON AND
PHOTO-FENTON CONVERSIONS (25 °C)
Pollutant conversion (%)
Conversion
CP/CF Fenton ε(%) photo-Fenton ε(%) Enhanc.(%)
Exp. Data
Predictions
Exp. Data
Predictions
Exp. Data
Predictions
3.3
3.3
5.4
5.4
8.4
8.4
29.3
31.1
37.6
39.7
43.2
45.7
6.1
5.6
5.8
80.7
81.0
80.6
80.2
79.3
78.6
0.4
0.5
0.9
175.4
160.4
114.4
102.0
83.6
72.0
N
A conversion of 81% has been achieved for the lowest CP/CF
N
The photo-Fenton system produces a conversion up to 175%
greater than that obtained with the Fenton reaction (CP/CF=3.3)
COMPARISON BETWEEN FENTON AND
PHOTO-FENTON CONVERSIONS (25 °C)
Pollutant conversion (%)
Conversion
CP/CF Fenton ε(%) photo-Fenton ε(%) Enhanc.(%)
Exp. Data
Predictions
Exp. Data
Predictions
Exp. Data
Predictions
3.3
3.3
5.4
5.4
8.4
8.4
29.3
31.1
37.6
39.7
43.2
45.7
6.1
5.6
5.8
80.7
81.0
80.6
80.2
79.3
78.6
0.4
0.5
0.9
175.4
160.4
114.4
102.0
83.6
72.0
Notice that the photo-Fenton conversion decreases
when the CP/CF initial molar ratio is increased.
EFFECTS OF THE H2O2 ON FORMIC
ACID CONVERSION (T = 25 °C)
The change in the H2O2 concentration (CP) may have
two opposite effects:
N
At low CP, ferrous ion (Fe2+) generation may be too
low and so will be the OH• generation.
N
At high CP, H2O2 acts as a radical trapping agent, thus
competing with the pollutant degradation path and
rendering lower degradation rates:
H2O2 + HO• → HO2• + H2O
N
Thus, an optimal molar ratio CP/CF should be expected.
PARAMETRIC STUDY: EFFECTS OF THE
H2O2 ON FORMIC ACID CONVERSION
XF (t = 1 h) vs. CP/CF:
‹ Fenton and ph-Fenton
‹ θZ = 10°, 40°, 70°
1.0
θ z = 10º
40º
0.8
X F (t = 1h)
Photo-Fenton
0.6
²
70º
0.4
Fenton
Fenton
CFe
= 0.46
CFe
3+/CC
F F= 0.46
2.2mM
mM
CFC=
F =2.2
3+
0.2
0.0
0
2
4
6
CP CF
8
²
10
At high values of θZ,
increasing the CP/CF
ratio increases the
conversion
At low values of θZ
(high radiation), an
optimal molar ratio
CP/CF is observed
COMPARISON BETWEEN FENTON AND
PHOTO-FENTON CONVERSIONS (t = 20 min)
XF (20 min)
100
N
80
60
N
40
20
N
0
20
30
40
50
Temperature (ºC)
60
Model predictions of
formic acid conversion:
¾ Fenton (
)
¾ photo-Fenton (
)
Experimental data:
¾ Fenton (L )
¾ photo-Fenton („)
Increasing the reaction
temperature decreases
the enhancement of the
pollutant conversion
POLLUTANT CONVERSION AND
CONVERSION ENHANCEMENT (t = 20 min)
N
XF (20 min)
Enhacement (%)
200
186 (%)
74 (%)
150
7 (%)
N
100
50
N
0
25
40
55
Temperature (ºC)
N
UV solar radiation
improves the
effectiveness of the
Fenton process
For the lowest
temperature 25°C, the
pollutant conversion is
significantly increased
Intermediate behavior
for 40°C
For the highest
temperature 55°C, this
effect is less important
FINAL REMARKS
Increased reaction temperature can enhance the
reaction rate of the Fenton and photo-Fenton processes.
However, at higher temperatures: (i) this conversion
enhancement is less important and (ii) the efficiency of
hydrogen peroxide declines: decomposition of H2O2
into oxygen and water (Malik and Saha, 2003).
It is possible to take andvantage of the natural
temperature of a wastewater at the end of the process
(in the textil industry: Rodríguez et al., 2002).
Possibility of a combined photochemically and
thermally enhanced Fenton process, using solar energy
(UV/Vis + IR photons: Sagawe et al., 2001).
THANKS
N
Dr. Rubén D. Piacentini, Grupo de Energía Solar, Instituto
de Física Rosario (IFIR), Rosario - Argentina
N
UNIVERSIDAD NACIONAL DEL LITORAL (UNL)
(National University of Litoral, Santa Fe - Argentina)
N
CONSEJO NACIONAL DE INVESTIGACIONES
CIENTIFICAS Y TECNICAS (CONICET) (National
Council for Science and Technology of Argentina)
N
AGENCIA NACIONAL DE PROMOCION CIENTIFICA
Y TECNOLOGICA (ANPCYT) (National Agency for the
Promotion of Science and Technology of Argentina)