water research 43 (2009) 33553364

Available at www.sciencedirect.com

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

Mathematical modelling of hydrogen sulphide emission and

removal in aerobic biofilters comprising chemical oxidation
Jane Meri Santosa,*, Evenilson Soprani Lopesb, Neyval Costa Reis Juniora,
Leandro Melo de Sac, Nigel John Horand
a

Departamento de Engenharia Ambiental, Universidade Federal do Esprito Santo, Av. Fernando Ferrari 514, 29.060-970 Vitoria, ES, Brazil
Companhia Vale do Rio Doce, Av. Dante Michelini 5500, Ed. Belesa Bloco II Superior, Ponta de Tubarao, 29.090-900 Vitoria, ES, Brazil
c
Centro Federal de Educacao Tecnologica do Esprito Santo, Av. Arino Gomes Leal 1700, 29.700-603 Colatina, ES, Brazil
d
School of Civil Engineering, The University of Leeds, Leeds LS2 9JT, UK
b

article info

abstract

Article history:

Four different empirical expressions have been compared for estimating the removal of

Received 9 May 2008

hydrogen sulphide (H2S) from wastewater by chemical oxidation during its treatment in an

Received in revised form

aerated biofilter. The relative importance of this removal process is considered in a mass

15 September 2008

balance proposed by an emission model. Two of the four models investigated were able to

Accepted 28 November 2008

predict the mean H2S removed fraction within a confidence interval of 95% and they

Published online 24 December 2008

demonstrated good agreement with experimental data. Biodegradation and oxidation were
the two main removal mechanisms in the biofilter whereas stripping and volatilization

Keywords:

made only minor contributions. However they can be of significance when the emission

Hydrogen sulphide

rates are calculated.

Odour emission

2008 Elsevier Ltd. All rights reserved.

Chemical oxidation
Aerobic biofilter
Emission model
Odour control

1.

Introduction

Areas of anaerobic microbial activity at domestic sewage

treatment works can generate unpleasant odours that may
be noticed in surrounding residential areas and may often
lead to complaints. Although modern wastewater treatment
plants (WWTP) have efficient odour control systems, where
they are close to densely populated areas then negative
impacts can still occur. This is a result of the efficiency of
the human olfactory system that is capable of detecting
odours at very low concentrations and at over very short
time intervals. Thus even very low emissions can generate

negative comments. In recent years complaints about

odours have increased due to the increasing number of
residences built in the vicinity of WWTP. As a result new
environmental legislation, or revision of existing legislation,
intended to control odour emission and its impact, has been
implemented in many countries including Germany, Austria,
UK, Australia, USA and Canada (Mahin, 2001). Hydrogen
sulphide (H2S) is the most common odourant gas emitted by
WWTP due both to its high emission rate and also its very
low detection threshold. As a result it is important to
understand the likely concentrations of this gas at the site
boundary of a WWTP and this is usually undertaken by

* Corresponding author. Tel./fax: 55 27 33352648.

E-mail addresses: janemeri@npd.ufes.br (J.M. Santos), evenilson.lopes@cvrd.com.br (E.S. Lopes), neyval@inf.ufes.br (N.C. Reis Junior),
salmelo@hotmail.com (L.M. de Sa), n.j.horan@leeds.ac.uk (N.J. Horan).
0043-1354/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2008.11.055

3356

water research 43 (2009) 33553364

dispersion modelling. However in order to apply a dispersion

model to evaluate odourant H2S concentration at the
boundaries of the WWTP it is necessary to estimate the gas
emission rate as this is needed as input data for the
dispersion model.
If the WWTP is in operation this offers the opportunity of
measuring emission rates. However, mathematical emission
models can offer an alternative option for estimating emission rates both for a planned WWTP that is not yet constructed, or as a low cost and rapid option for those already
operational. Mathematical models are developed to estimate
the emission rates for each unit process that comprises the
WWTP flow train. There are several mathematical models
available in the literature, such as WATER9 (USEPA, 2001),
TOXCHEM (Enviromega, 2003), Gostelow et al. (2001), which
are based on a mass balance of the odorant compound dissolved in the liquid phase with all the removal mechanisms
(physical, chemical or biological) described by empirical
algebraic expressions. The models generally consider that
the main mechanisms of removal of an odorant compound
from the liquid phase are volatilization, stripping, absorption, adsorption and biodegradation. The exact mechanism
will be a function of the engineered design of each unit
process.
Santos et al. (2006) presented a performance evaluation of
four different mathematical models to estimate the removal
of hydrogen sulphide for a WWTP with an UASB reactor followed by aerobic biofilters, namely AP-42 (USEPA, 1995),
WATER8 (USEPA, 1994), TOXCHEM (Enviromega, 2003) and
Gostelow et al. (2001). The models were evaluated by
comparing measured H2S removal rates with experimental
data. An analysis of the performance of the models for each
unit of the WWTP indicated that all the models were able to
estimate the H2S concentrations and removal rates within the
95% confidence interval, except for aerobic biofilters. In this
case, the selected models significantly underestimated the
H2S removal from the liquid phase, which may suggest that
not all removal mechanisms are correctly represented by the
models tested.
In aerobic processes, chemical oxidation can also be of
importance as a removal process since the oxygen dissolved
in the liquid phase can oxidize the odorant compound (dissolved hydrogen sulphide and others). As models of odorant
gas emission are based on a mass balance of the odorant
compound then chemical oxidation mechanisms may work
as a sink for the compound in the mass balance. The emission models found in the literature (for instance, WATER9
(USEPA, 2001), TOXCHEM (Enviromega, 2003), Gostelow
et al. (2001)) do not include oxidation as a removal mechanism. This paper intends to compare four different empirical
expressions for estimating the chemical oxidation removal of
hydrogen sulphide (H2S) from sewage in an aerated biofilter
and to indicate the relative importance of the oxidation
removal process. This work proposes to include the oxidation process as a H2S removal mechanism in the emission
model proposed by Gostelow et al. (2001) (described in details
in Section 2), which according to Santos et al. (2006) showed
the best agreement with the experimental data when
compared to other emission models available in the
literature.

2.

General fate model

A number of mathematical emission models for volatile

organic compounds (VOC) and H2S emitted by WWTP are
available both freely and commercially. An important
assumption of these models is that each unit of the WWTP is
treated as an homogeneous reactor, regardless of the fluid flow
and mass transfer that is responsible for the concentration
distribution of the pollutant inside these units. The pollutant
can accumulate, pass through the reactor or be affected by one
or more of the following removal mechanisms: volatilization
from the sewageatmosphere interface, stripping due to
diffusion into air bubbles, biodegradation (which is also
a source of hydrogen sulphide under anaerobic conditions),
adsorption of solid particles or biomass, absorption for insoluble liquids and chemical oxidation for aerated reactors.
Thus, assuming homogeneous mixing, the mass balance
can be described as presented by Corsi and Olson (1998) except
that an extra term (Rq) has been included to account for
chemical oxidation as a removal mechanism:
V

dC
QCo  QC Rv Rs Rb Rad Rab Rq
dt

(1)

where Co is the contaminant concentration in the liquid phase

entering the unit (g m3); C is the contaminant concentration
after it has been mixed within the fluid inside the unit; V is the
volume of the unit (m3), and Q is the sewage flow rate (m3 s1).
Rv, Rs, Rb, Rad, Rab and Rq represent the contaminant removal
rates (g s1) by volatilization, stripping, biodegradation,
adsorption, absorption and chemical oxidation, respectively.
Sulphide is present in domestic sewage as dissolved sulphide
(H2S HS S2) and trace metal sulphide (e.g., ferrous
sulfide, FeS). However, the contaminant concentration in
equation (1) is only related to the non-ionized form of sulphide
(H2S) which is the one that is actually emitted to the
atmosphere.
The algebraic equations that give the removal rates by
volatilization, stripping and biodegradation processes presented in this paper are as those proposed by Gostelow et al.
(2001) and shown by Santos et al. (2006) to give the best results
when compared to other empirical equations. Absorption was
not included in the model as a removal mechanism of H2S
from domestic sewage since it is assumed not to contain
immiscible liquids such as oils (which would occur in industrial sewage). Adsorption of H2S can occur in the sludge inside
the reactor, however, it is supposed to occur very slowly after
the reactor reaches a steady regime and thus adsorption can
be assumed negligible as a H2S removal mechanism for a time
frame to which the model and measurements were applied.

2.1.

Volatilization from quiescent surfaces

Volatilization is defined as the transfer of a mass of compound

through the gasliquid interface. The rate of removal of
a pollutant by volatilization is modelled according to Corsi and
Olson (1998) as


Cg
A
Rv KL C 
Hc

(2)

3357

water research 43 (2009) 33553364

where KL is an overall mass transfer coefficient (m s1), A is the

interfacial area over which the mass transfer occurs (m2), the
concentration Cg represents the contaminant concentration
(g m3) in the gaseous phase away from the liquid surface in
which case for a open reactor it is reasonable to assume that
Cg y0, i.e., the background concentration in the atmosphere is
zero, Hc is the Henry non-dimensional constant for the
contaminant (m3liq m3
gas) given as
Hc

H
RT

(3)

where H is the Henry constant (Pa m3 mol1), R is the universal

gas constant (8.31 Pa m3 mol1 K1) and T is the wastewater
temperature (K), then,
Rv KL AC

(4)

The overall mass transfer coefficient (KL) for quiescent

surfaces can be deduced from the two-film theory as:
1
1
1

KL kL Hc kG

kL 0:0035U ScL 0:5

0:67

kG 0:04U ScG

(6)

kv Uz

lnzz0

(8)

where kv is the Von Karman constant (non-dimensional), Uz is

the wind velocity (m s1) at a height z (m) and z0 is the
roughness parameter (m).

2.2.

Stripping

In units with aeration, bubbles formed by air injection in the

liquid phase carry the contaminant and release it to the
atmosphere after reaching the free surface. The removal rate
through air bubbles spreading in the sewage can be modelled
as
Rs Qb CHc g

where a is the interfacial area per volume unit (m2 m3) of the
aerated tank and V is the tank volume (m3) occupied by the
sewage. The overall mass transfer coefficient (KL) for different
pollutants are related to the coefficients of transfer of mass of
the oxygen by an expression given by Munz and Roberts
(1989):
KL;C

KL;O2

DL;C
DL;O2

n 
1

1
Hc kG =kL

1
(11)

where DL,C and DL ;O2 are the coefficients of diffusion in the

liquid phase for the odorant compound and for the oxygen,
respectively. The value of the exponent n is considered to be
0.6 (Corsi et al., 1992). The ratio kG/kL (between 2.2 and 3.6) is
determined from experimental studies (Hsieh et al., 1993,
1994).

2.3.

Biodegradation

Oyarzun et al. (2003) described the biodegradation of H2S by

bacteria of the genus Thiobacillus according to the reactions
below which use the ionized form of sulphide (HS) and
generate a non-odorant compound

2HS O2 / 2S0 2OH

(12)

2S0 3O2 / 2SO2

4 2H

(13)

(7)

where ScL and ScG are the Schmidt numbers for the contaminant in the liquid and gas phase, respectively.
Friction velocity can be calculated using the vertical profile
of wind velocity obtained from the similarity theory for
neutral conditions:
U

(10)

(5)

Wind flow conditions over quiescent surfaces strongly

affect mass transfer coefficients for the liquid and gas phases
(kL and kG in m s1) due to turbulence and wave generation.
Friction velocity (U* in m s1) is used to represent atmospheric
wind flow conditions. The formulations used to calculate kL
and kG are those presented by Mackay and Yeun (1983).




KL aV
g 1  exp
Qb Hc

(9)

where Qb is the bubble volumetric flow rate (m3 s1), g

represents the degree of saturation (chemical equilibrium)
achieved by a bubble by the time it reaches the sewageair
interface. Matter-Muller et al. (1981) considered that the
bubbles were partially saturated and the expression presented below can be used to estimate the degree of
saturation.

The biological degradation kinetics is represented by the

well-known MichaelisMenten kinetics developed in 1913 to
describe enzymatic reactions with a single substrate (equation (14)). Thus, the biodegradation removal process is
modelled as
Kmax bi VC
Rb 
Ks C

(14)

where Kmax is the maximum biorate constant (g s1 g1 of

biomass), bi is the active biomass concentration in the reactor
(g m3) and Ks is the half-saturation biorate constant (g m3)
(Van Langenhove and De heyder, 2001). For low concentrations, it can be simplified by assuming Ks [ C, such that the
biodegradation removal rate becomes
Rb Kb bi VC

(15)

where Kb is the first order biodegradation constant rate

(m3 g1 s1) calculated as the rate between Kmax and Ks.
The pH value and temperature are important parameters
in biological degradation kinetics. Li et al. (1998) observed
maximal removal of odorous compounds in a biofilter under
mesophilic conditions growing between pH values of 6.0 and
9.0, whereas Pronk et al. (1990) noted that inorganic sulphide
compounds are oxidised by predominantly acidophilic
bacteria such as Thiobacilli and which are often able to grow
thermophilically.

3358

3.

water research 43 (2009) 33553364

Chemical oxidation

Several works have been published that consider chemical

oxidation of sulphide in pure water, sea water and sewage.
The results obtained vary from author to author which can be
explained by differences in concentration of the sulphide ion,
dissolved oxygen, pH, temperature, catalysis for metals and
the applied methodologies for determination of the
compounds in each study (Chen and Morris, 1972; Kuhn et al.,
1983; Millero et al., 1987; Wilmot et al., 1988; Zhang and
Millero, 1993).
Assuming that the sewage has a homogenous composition, the temperature is constant and that there is no other
substances affecting the reaction, Mortimer (2000) presented
chemical oxidation rate of H2S written as:
Rq dC

kCm O2 n
V
dt

(16)

where C and [O2] represent hydrogen sulphide concentration

and dissolved oxygen concentrations, respectively. The
constant of proportionality depends on temperature and
pressure, although the dependence on pressure can be
assumed negligible (Mortimer, 2000).
Chen and Morris (1972) evaluated the parameters k, m and
n in equation (16) to determine the hydrogen sulphide removal
rate by chemical oxidation in a sulphide solution containing
sodium chloride to maintain a steady pH. The authors
obtained values for m and n as 1.34 and 1.56, respectively and
for k at pH 6.9, they obtained a value of 4.7 (mol L1)0.9 h1.
From their experiments, Chen and Morris (1972) also verified
the influence of pH on the chemical removal rate. The rate
increases from pH 6.0 reaching a maximum value at pH 8.5
and decreasing to a minimum at pH 9.3 and increasing again

Grit
Chamber

reaching a second maximum value at pH 11.5. They explained

that this behaviour is due to dissociation of molecular
hydrogen sulphide and formation of polysulphides.
Jolley and Forster (1985) carried out a study into sulphide
oxidation in pure water and sewage and obtained values for m
and n as 0.82 and 1.19, respectively. For k, they obtained
a value of equal to 66.2 (mol L1)1 min1.
Wilmot et al. (1988) determined the rate of sulphide
oxidation at three different large WWTPs in Perth, Australia.
The sulphide concentration in domestic sewage varied from
0.2 to 8.0 mg L1 and the oxygen concentration from 5.0 to
20.0 mg L1. From this study they obtained values for m and n
as 0.38 and 0.21, respectively and as for k, they obtained
a value of 0.055 min1.
Buisman et al. (1990) reviewed the chemical oxidation of
the sulphide in pure water, sewage and seawater. The authors
presented five equations to describe the chemical oxidation
rate of sulphide and the constants n and k found for each one
of the five equations were obtained based on four experiments. During these experiments, Buisman et al. (1990) varied
the dissolved oxygen concentration between 0.1 and
8.5 mg L1 and the hydrogen sulphide concentration between
5.0 and 300.0 mg L1. They found the removal rate to vary from
0.008 to 1.000 mg min1 L1 of sewage and equation (17) was
shown to best estimate the oxidation rate
Rq
kS0:41 O2 0:39 logS
V

where the value for k was 0.57 mgp Lq.

Nielsen et al. (2004) investigated the kinetics and stoichiometry of chemical sulphide oxidation in sewers. The effect of
pH and temperature was investigated over the range 69 and
525  C, respectively. The stoichiometry of the chemical
oxidation was not significantly affected by varying pH and

B
Biogas

(17)

Distribution
box

BF
Outlet

Biofilter
5

UA SB

4
3

Sewage

BF
Sampling
points

Laundering sludge

Aerator

Treated
effluent

Fig. 1 Schematic representation of the experimental WWTP. The points A (grit chamber inlet), B (distribution box inlet) and
C (BF outlet) are the sampling points for measurements of sulphide and sulphate concentrations, and points 15 represent
the sampling points in the biofilter for measurements of DO concentration.

3359

water research 43 (2009) 33553364

temperature. Based on experiments, a general rate equation

(equation (18)) including a stoichiometry coefficient
describing chemical sulfide oxidation in varying pH and
temperature conditions was proposed, with m and n equal to
0.9 and 0.2 respectively. The k constant rate reaction was
determined as function of pH and temperature, in accordance
with equation (18):
 
k0 k1 K1 = H
Rq
  S0:9 O2 0:2 1:06T20

V
1 K 1 = H

4.1.
Measurements of H2S and SO2
4 concentration in the
liquid phase

(18)

where k0 is the constant for oxidation of H2S equal to

1
0:04gS m3 0:1 gO2 m3 0:2 h , k1 is the constant for oxidation
1

3 0:1
of HS equal to 0:5gS m gO2 m3 0:2 h , K1 is the first
dissociation constant for H2S taken as 1.0  107 (nondimensional) and T is Celsius temperature. The k value varies
between 9.30  103 and 1:33mgS L1 1m mgO2 L1 n h1 .
It should be noted that it is rather confusing the unit of the
constant rate reaction k presented in the investigated literature.
Different units are presented which are not consistent with the
rate equation. The work presented by Nielsen et al. (2004) seems
to be the one which presents the units in a consistent manner.
Chen and Morris (1972) have presented it in mol0.9 h1 which
implies that the authors made no difference between sulphide
concentration unit which should be molS L1 and dissolved
oxygen concentration molO2 L1 by multiplying the molar
mass of sulphide and dissolved oxygen to obtain a single power.
Thus, for the rate equation to be consistent the constant rate
should have been given as 4:7molS L1 1m molO2 L1 n h1 or
4:7molS L1 0:34 molO2 L1 1:56 h1 . Similarly, Jolley and Forster (1985) should have presented the rate constant as
66:2molS L1 0:18 molO2 L1 1:19 h1 . Wilmot et al. (1988) presented a value of 0.055 min1 to indicate the rate constant
or
instead
of
0:055mgS L1 0:62 mgO2 L1 0:21 min1
1 0:62
3:0molS L molO2 L1 0:21 h1 . The model presented by
Buisman et al. (1990) gives an undefined unit for k as it involves
a power written as a logarithm of a dimensional quantity (the
sulphide concentration), in any case it should include a time
dimension in order to make the rate equation consistent.

4.

achieved. Nitrification was not observed on the bed as judged

by the absence of nitrate in the effluent.
The experiments were carried out between 23rd June and
17th October 2004 and all sampling was undertaken between
13:00 and 14:00 hour.

Experimental work

An experimental WWTP located at the Federal University of

Espirito Santo (UFES) in Brazil was used for the experiments
reported in this work. This WWTP treats sewage from a population
of 1000 people with a daily flow rate of 86 m3. Treatment
comprises a pumping station, screening and grit-removal, upflow anaerobic sludge blanket reactor (UASB), three secondary
biofilters (BF) and one tertiary biofilter, UV disinfection and
sludge drying beds (Fig. 1).
The three secondary biofilters (BF1, BF2 and BF3) and the
tertiary biofilter (BF4) had a combined volume of 6.3 m3 and
specific surface of 465, 459, 1234 and 867 m3 m2, respectively.
BF1, BF2 and BF3 were fed with effluent from the UASB which
had a BOD/COD of 0.45 at a hydraulic loading rate of
53.2 m3 m2 d each and BF4 received the effluent from the
three secondary biofilters (at a hydraulic loading rate three
times higher). The average bed temperature was 23.4  C and
under these conditions a BOD/COD removal of 24% was

Samples were collected at the inlet and outlet of BF in order to

evaluate its removal efficiency, as indicated in Fig. 1. Eighteen
experimental campaigns were carried out to measure
concentrations of H2S. At each sampling point, two samples
were taken for H2S analysis and three samples for sulphate
analysis. H2S in wastewater was determined by the iodometric method a preliminary determination of the total
sulphide (S2) concentration and for subsequent determination of the non-ionized hydrogen sulphide (H2S) concentration. For analysis of SO2
4 concentration both turbidimetry and
gravimetry with residue drying were employed. All procedures were as described by APHA (1995).

Table 1 Operational parameters, meteorological data

and physicalchemical constants.
Parameters

Unit
3

1

Sewage flow rate

m s
Air flow
m3 h1
rate (aeration in BF)
Overall mass
m s1
transfer coefficient
for O2 (in BF)
Active biomass
g L1
concentration (in BF)
L h1 g1
First order
biorate constant for H2S
atm m3 mol1
Henrys coefficient
for H2S
cm2 s1
Diffusion coefficient
for H2S in water
cm2 s1
Diffusion coefficient
for H2S in air
cm2 s1
Diffusion coefficient
for O2 in water
Water specific mass
g cm3
Air specific mass
g cm3
Water dynamic viscosity
g cm1 s1
Air dynamic viscosity
g cm1 s1
Roughness length
m
Mean wind
m s1
velocity at 10 m height

Mean ambient
C
air temperature

Symbol

Value

Q
Qb

3.333  104a
3.0a

KL;O2

1.0  104b

bi

15.0c

Kb

0.3768d

0.023d

DL

1.61  105d

DG

0.176d

DL;O2

2.4  105e

rL
rG
mL
mG
z0
Uz

1.0e
1.2  103e
8.4  103e
1.8  104e
1.0f
2.7g

Tc

26.7g

a Values taken from measurements.

b Experimental values obtained by Silva (2003) in pilot aerated BF.
c Value assumed to be equal at 0.75  TS (total solids), estimated in
20 g L1.
d USEPA (1994).
e USEPA (1995).
f Seinfeld and Pandis (1998).
g Mean value obtained from two hourly means (measured one
hour before starting each campaign and one hour after).

3360

water research 43 (2009) 33553364

L1
Table 2 Observed mean concentration of SO2L
).
4 and H2S in the liquid phase (mg L

Date

Trial

06/23/04
06/24/04
06/25/04
06/26/04
06/27/04
06/29/04
06/30/04
07/01/04
07/02/04
07/07/04
07/08/04
07/09/04
07/10/04
10/09/04
10/10/04
10/14/04
10/15/04
10/17/04
Mean  uncertainty
Mean  95% CI

4.2.

Sulphate (mg L1)

Hydrogen sulphide
(mg L1)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

BF inlet

BF outlet

BF inlet

BF outlet

5.10  0.85
4.99  0.93
4.67  0.52
4.63  0.47
3.93  0.22
6.51  0.63
7.84  0.57
5.73  0.34
8.85  0.91
8.23  0.91
5.05  0.88
3.54  0.39
2.83  0.34
5.73  0.65
4.65  0.53
6.00  0.39
5.87  0.54
6.44  0.68
5.59  0.15
5.59  0.79

0.58  0.37
0.56  0.41
0.49  0.34
0.47  0.25
0.44  0.12
0.36  0.28
0.45  0.15
0.42  0.13
0.37  0.22
0.82  0.31
0.49  0.40
0.32  0.17
0.44  0.14
0.64  0.32
0.44  0.21
0.49  0.13
0.65  0.15
0.53  0.13
0.50  0.06
0.50  0.06

7.50  0.21
3.24  0.14
6.66  0.19
6.66  0.28
1.91  0.08
4.60  0.19
8.79  0.25
5.34  0.22
12.44  0.52
19.09  0.80
15.65  0.66
13.75  0.58
7.27  0.31
6.69  0.40
13.28  0.42
15.35  0.08
17.53  0.23
9.56  0.18
9.74  0.09
9.74  2.53

39.42  1.66
31.60  1.33
25.14  1.06
25.16  1.06
24.03  1.01
35.33  1.48
41.66  1.52
30.55  1.28
54.90  2.31
63.47  2.67
35.18  1.48
37.85  1.59
43.82  1.84
30.08  1.12
30.69  1.19
30.08  0.83
31.56  0.82
33.64  0.74
35.79  0.35
35.79  5.08

Measurement of dissolved oxygen concentrations

4.3.

Dissolved oxygen concentration was analysed using a Digimed DM4 oxygen meter. Ten samples (at intervals of 2 min)
were taken from each of the five points shown in Fig. 1 to
produce a vertical profile of dissolved oxygen inside the biofilter (for each campaign).

Measurements of emission rate

For sampling the gas phase the apparatus described by Bliss

et al. (1995) was used. Concentration measurements in the gas
phase were carried out with an electrochemical analyzer
manufactured by Interscan Corporation (model portable
series 4000). The emission rate was calculated as the gas

Table 3 Observed vertical profile of mean dissolved oxygen concentration in the BF.
Date

Trial

Observed dissolved
oxygen concentration (mg L1)
Sampling points

06/23/04
06/24/04
06/25/04
06/26/04
06/27/04
06/29/04
06/30/04
07/01/04
07/02/04
07/07/04
07/08/04
07/09/04
07/10/04
10/09/04
10/10/04
10/14/04
10/15/04
10/17/04
Mean  uncertainty
Mean  95% CI

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

0.81  0.15
0.61  0.15
0.62  0.15
0.71  0.15
0.82  0.15
0.50  0.15
0.49  0.15
0.54  0.15
0.90  0.15
1.14  0.16
0.66  0.15
0.95  0.16
0.82  0.15
0.61  0.15
0.62  0.15
0.30  0.14
0.50  0.15
0.48  0.15
0.67  0.04
0.67  0.10

3.39  0.20
2.53  0.19
2.59  0.19
2.96  0.20
3.41  0.20
2.10  0.18
3.39  0.20
3.35  0.20
2.41  0.18
2.40  0.18
2.38  0.18
2.98  0.20
2.87  0.19
2.87  0.19
1.28  0.16
0.78  0.15
0.72  0.15
0.91  0.15
2.41  0.04
2.41  0.45

6.95  0.28
5.18  0.24
5.29  0.24
6.07  0.26
6.98  0.28
4.30  0.22
4.73  0.23
4.71  0.23
3.63  0.21
4.26  0.22
4.55  0.23
4.91  0.23
4.86  0.23
4.86  0.23
4.09  0.22
4.35  0.22
2.21  0.18
4.60  0.23
4.81  0.05
4.81  0.55

4.84  0.23
3.61  0.21
3.69  0.21
4.22  0.22
4.86  0.23
3.00  0.20
2.61  0.19
2.63  0.19
2.79  0.19
3.22  0.20
3.78  0.21
3.25  0.20
3.71  0.21
3.71  0.21
3.22  0.20
2.69  0.19
1.69  0.17
2.08  0.18
3.31  0.05
3.31  0.42

6.30  0.26
4.70  0.23
4.80  0.23
5.50  0.25
6.33  0.26
3.90  0.21
4.53  0.23
4.54  0.23
4.30  0.22
4.40  0.22
3.73  0.21
3.39  0.20
3.68  0.21
3.68  0.21
4.35  0.22
3.70  0.21
3.46  0.21
3.48  0.21
4.38  0.05
4.38  0.45

3361

water research 43 (2009) 33553364

concentration (Cg) multiplied by the volume of air flow

through the apparatus for a unit of time. Gas phase
measurements were available for only two of the 18
campaigns, as the H2S concentrations were below the detection range of the instrument used. Measurements were taken
at intervals of 30 s and a total of 85 samples were taken for
each of the two campaigns. The air flow rate inside the wind
tunnel was 0.5 m3 s1 and consequently the air velocity was
1.7 m s1, which is similar to the wind speed in the area.

A direct correlation between the friction velocity and the

measured wind velocity was obtained by assuming a logarithmic velocity profile with surface roughness size (z0) equal
to 1.0, as suggested by Seinfeld and Pandis (1998) for regions
with trees. The mean ambient air temperature, mean wind
speed and relative humidity were 26.7  1.7  C, 2.7  0.8 m s1
and 64.1  6.0%, respectively (within 95% confidence interval).

4.5.

Modelling setup

In order to run the emission model certain parameters were

measured and others were taken from the literature as an
approximation of the actual value as shown in Table 1.

Predicted concentration (mg/L)

The hydrogen sulphide concentration showed a large

decrease across the biofilter, whereas there was a larger
increase in the concentration of sulphate (Table 2).

5.2.

Measurements of meteorological data

0.80

0.60

0.40

0.20
JOLLEY & FORSTER (1985)

0.00
0.00

0.20

0.40

0.60

0.80

1.00

0.80

0.60

0.40

0.20
WILMOT et al. (1988)

0.00
0.00

1.00

Observed concentration (mg/L)

0.20

0.40

0.60

0.80

1.00

Observed concentration (mg/L)

D 1.00

1.00

0.80

0.60

0.40

0.20
BUISMAN et al. (1990)

0.00
0.00

0.20

0.40

0.60

0.80

1.00

Observed concentration (mg/L)

Predicted concentration (mg/L)

Predicted concentration (mg/L)

Measurements of dissolved oxygen (OD)

Table 3 shows the observed mean concentrations of dissolved

oxygen along the vertical profile of BF. Sampling points 15 are
indicated in Fig. 1. In the first point at the bottom of BF, there is
low DO concentration which characterizes the UASB outflow.
The second point is located above the aeration duct and on the
base of the fixed bed, thus the average concentration of DO
increases in relation to that observed at point 1. The third
point is located in the middle of the fixed bed and above the
aeration duct and shows an even higher average concentration of OD reaching the maximum value of 4.81 mg L1. The
fourth point is located in the higher part of the fixed bed where
the average concentration of DO decreases to 3.31 mg L1 due
to the consumption of oxygen. Finally, the fifth point is located

1.00

Results and discussion

5.1.
Concentration measurements of hydrogen sulphide
and sulphate

Predicted concentration (mg/L)

4.4.

5.

0.80

0.60

0.40

0.20
NIELSEN et al. (2004)

0.00
0.00

0.20

0.40

0.60

0.80

1.00

Observed concentration (mg/L)

Fig. 2 H2S concentration results predicted by the model implemented with four formulations for chemical oxidation: (A)
Jolley and Forster (1985), (B) Wilmot et al. (1988), (C) Buisman et al. (1990), (D) Nielsen et al. (2004) versus H2S measurements
concentrations.

3362

water research 43 (2009) 33553364

Table 4 Observed and predicted values of mean removal rate and removed fraction of hydrogen sulphide in the BF unit.
Jolley and Forster (1985) Wilmot et al. (1988) Buisman et al. (1990) Nielsen et al. (2004)
Predicted values Removal rate
(mg s1)
Removed fraction (%)
Observed values Removal rate
(mg s1)
Removed fraction (%)

1667.8

1774.9

1670.6

1659.5

89.6
1696.7  257.5a

95.5

89.7

89.1

90.6  1.3a

a 95% confidence interval.

at the sewageair interface and shows an increase in DO

concentration which could be explained by its proximity to
the atmosphere.
For wastewater, the chemical sulphide oxidation has been
reported to be slow when either reactant is present in
concentrations of less than 1 g m3 (ASCE, 1989). And, as
presented in Table 3, DO is significantly above 1 g m3, except
at the bottom of the biofilter just below the oxygen injection,
which means that oxygen diffusion was not a limiting factor
of the removal process.

5.3.
Intercomparison of chemical oxidation
removal models
Fig. 2 shows the effluent concentrations for H2S observed and
predicted for models in the biofilter. The concentrations were
calculated using the model described by Equation (1) including
the removal of H2S by volatilization (Equation (4)), stripping
(Equation (9)), biodegradation (Equation (15)) and chemical
oxidation (Equations (16), (17) or (18)). Of the four equations
described in the literature for the mechanism of chemical
oxidation, namely: Jolley and Forster (1985), Wilmot et al.
(1988), Buisman et al. (1990) and Nielsen et al. (2004); the
equations proposed by Jolley and Forster (1985), Buisman et al.
(1990) and Nielsen et al. (2004) gave the best results. The model
proposed by Wilmot et al. (1988) significantly underpredicted
the H2S concentrations at the outlet of BF.

5.4.

Removal and emission rates

The observed and predicted removal rates shown in Table 4 indicate

that the models proposed by Wilmot et al. (1988) and Nielsen et al.

(2004) were not able to predict the mean removed fraction

within the 95% confidence interval.
Fig. 3 presents the predicted and observed removed fractions of H2S for all trials. In general, the model proposed by
Wilmot et al. (1988) has overestimated the H2S removal,
although at three of the 18 trials the model showed a good
agreement with the experimental results. By contrast, the
results obtained from the models proposed by Jolley and
Forster (1985), Buisman et al. (1990) and Nielsen et al. (2004)
showed good agreement with the experimental data in 13
trials. For one trial (07/10/04), all models significantly overpredicted the removed fraction.
Fig. 4 shows the observed and modelled emission rates for
H2S. The emission rates predicted by Jolley and Forster (1985),
Buisman et al. (1990) and Nielsen et al. (2004) were significantly and consistently higher than those predicted by Wilmot et al. (1988). Measurements of H2S emission rate were
conducted during two trials as shown in Fig. 4, the results
show better agreement with the results obtained by the model
proposed by Wilmot et al. (1988), however, no conclusion can
be drawn for these results since measurements were carried
out at only 2 of the 18 trials.

5.5.
Contribution of chemical oxidation to the overall
removal of hydrogen sulphide
Fig. 5 presents the predicted mean removal fraction for each
removal mechanism included in the models (biodegradation,
stripping, volatilization and oxidation). Biodegradation was
found to be the main mechanism responsible for the H2S
removal from the liquid phase (85%) when chemical oxidation

120.0

H2S emission (g/s)

Removed fraction (%)

100.0

95.0

90.0

85.0

100.0
80.0
60.0
40.0
20.0
0.0

80.0
1

9 10 11 12 13 14 15 16 17 18

NIELSEN et al. (2004)

WILMOT et al. (1988)

BUISMAN et al. (1990)

9 10 11 12 13 14 15 16 17 18

Experiment

Experiment
JOLLEY & FORSTER (1985)

JOLLEY & FORSTER (1985)

NIELSEN et al. (2004)

WILMOT et al. (1988)

Observed

BUISMAN et al. (1990)

Observed

Fig. 3 Observed and modelled removed fractions of H2S.

Fig. 4 Observed (trials 15 and 18) and predicted values of

H2S emission rates (all trials).

Predicted removal contribution (%)

water research 43 (2009) 33553364

(1988) with oxidation being the most important one. Stripping

and volatilization made only minimal contributions to H2S
removal in the BF, although they may still be of significance
when the emission rates are calculated due to the low
concentration at which H2S can cause nuisance.

100
80

JOLLEY & FORSTER (1985)

WILMOT et al. (1988)
BUISMAN et al. (1990)
NIELSEN et al. (2004)

60
40

references
20
0
Vol

Strip

Bio

Oxi

Fig. 5 Medium percentile results, obtained for the

contributions of the mechanisms of chemical oxidation,
biodegradation, stripping and volatilization for removal of
H2S of the liquid phase.

was accounted as proposed by Jolley and Forster (1985), Buisman et al. (1990) and Nielsen et al. (2004). However biodegradation was predicted as the second most important
mechanism (37%) when oxidation was accounted as proposed
by Wilmot et al. (1988), and in this last case, oxidation was
predicted as the main removal mechanism (60%).
The results obtained indicated that stripping and volatilization are responsible for removing less than 3.1% and 1.4%
respectively of the H2S. Although, the contribution of volatilization and stripping in the removal process of H2S from the
liquid phase is small, those two mechanisms are the ones to
account for when calculating the actual emission rate and
a very low H2S concentration may still be able to cause
nuisance.

6.

3363

Conclusion

This work intended to compare the use of four different

empirical expressions proposed by Jolley and Forster (1985),
Buisman et al. (1990), Wilmot et al. (1988) and Nielsen et al.
(2004) for estimating removal of hydrogen sulphide (H2S) by
chemical oxidation from sewage treated in a aerated biofilter
and to present the relative importance of this removal process
in the mass balance proposed by an emission model. The
models of Jolley and Forster (1985) and Buisman et al. (1990)
were able to predict the mean removed fraction of H2S within
a confidence interval of 95%. The predicted results from Jolley
and Forster (1985), Buisman et al. (1990) and Nielsen et al.
(2004) models showed better agreement with the experiments
than those predicted by Wilmot et al. (1988) which has, in
general, overpredicted the removed fraction.
Biodegradation and oxidation were found to be the two
main removal mechanisms in BF. Biodegradation was the
main mechanism responsible for the H2S removal from the
liquid phase when chemical oxidation was considered as
suggested by Jolley and Forster (1985), Buisman et al. (1990)
and Nielsen et al. (2004). However, biodegradation was predicted to be the second most important mechanism when
oxidation was accounted for as proposed by Wilmot et al.

APHA, 1995. Standard Methods for the Examination of Water and

Wastewater, 19th ed. American Public Health Association/
American Water Works Association/Water Environment
Federation, USA.
ASCE, 1989. American Society of Civil Engineers. Sulfide in
Wastewater Collection and Treatment Systems Manuals and
Reports on Engineering Practice 69, New York, USA.
Bliss, P.J., Jiang, K., Schulz, T.J., 1995. The development of
sampling system for determining odor emission rates from
areal surfaces: part II. Mathematical model. Journal of the Air
and Waste Management Association 45 (11), 989994.
Buisman, C., Ijspeert, P., Janssen, A., Lettinga, G., 1990. Kinetics of
chemical and biological sulphide oxidation in aqueous
solutions. Water Research 24 (5), 667671.
Chen, K.Y., Morris, J.C., 1972. Kinetics of oxidation of aqueous
sulfide by O2. Environmental Science and Technology 35 (6),
529537.
Corsi, R.L., Chang, D.P.Y., Schroeder, E.D., 1992. A modeling
approach for VOC emissions from sewers. Water Environment
Research 64 (5), 734741.
Corsi, R.L., Olson, D.A., 1998. Emission models. In: Rafson, H.J. (Ed.),
Odor and VOC Control Handbook. McGraw-Hill, USA, pp.
5.115.25.
ENVIROMEGA, 2003. Predicting VOC Emissions for Wastewater
Processing Using General Fate Models (GFM). http://www.
enviromenga.com/toxplus3.htm (acessed 16.01.03).
Gostelow, P., Parsons, S.A., Cobb, J., 2001. Development of an
odorant emission model for sewage treatment works. Water
Science and Technology 44 (9), 181188.
Hsieh, C.C., Babcock, R.W., Stenstrom, M.K., 1993. Estimating
emissions of 20 VOCs: 2. Diffused aeration. Journal of
Environmental Engineering, ASCE 119 (6), 10991118.
Hsieh, C.C., Babcock, R.W., Stenstrom, M.K., 1994. Estimating
semivolatile organic compound emission rates and oxygen
transfer coefficients in diffused aeration. Water Environment
Research 66 (3), 206210.
Jolley, R.A., Forster, C.F., 1985. The kinetics sulphide oxidation.
Environmental Technology Letters 6, 110.
Kuhn, A.T., Kelsall, G.H., Chana, M.S., 1983. A review of the air
oxidation of aqueous sulphide solutions. Journal of Chemical
Technology and Biotechnology 33A (8), 406414.
Li, X.Z., Wu, J.S., Sun, D.L., 1998. Hydrogen sulphide and volatile
fatty acid removal from foul air in a fibrous bed bioreactor.
Water Science and Technology 38 (3), 323329.
Mackay, D., Yeun, A.T.K., 1983. Mass-transfer coefficient
correlations for volatilization of organic solutes from water.
Environmental Science and Technology 17 (4), 211217.
Mahin, T.D., 2001. Comparison of different approaches used to
regulate odours around the world. Water Science and
Technology 44 (9), 87102.
Matter-Muller, C., Gujer, W., Giger, W., 1981. Transfer of volatile
substances from water to the atmosphere. Water Research 15
(11), 12711279.
Millero, F.J., Izaguirre, M., Sharma, V.K., 1987. The effect of ionic
interaction on the rates of oxidation in natural waters. Marine
Chemistry 22 (24), 179191.
Mortimer, R.G., 2000. Physical Chemistry. Elsevier, USA.

3364

water research 43 (2009) 33553364

Munz, C., Roberts, P.V., 1989. Gas- and liquid-phase mass transfer
resistances of organic compounds during mechanical surface
aeration. Water Research 23 (5), 589601.
Nielsen, A.H., Vollertsen, J., Hvitved-Jacobsen, T., 2004.
Chemical sulfide oxidation of wastewater effects of pH
and temperature. Water Science and Technology 50 (4),
185192.
Oyarzun, P., Arancibia, F., Canales, C., Aroca, G.E., 2003.
Biofiltration of high concentration of hydrogen sulphide using
Thiobacillus thioparus. Process Biochemistry 39 (2), 165170.
Pronk, J.T., Meulenberg, R., Hazeu, W., Bos, P., Kuenen, J.G., 1990.
Oxidation of reduced inorganic sulphur compounds by
acidophilic Thiobacilli. FEMS Microbial Reviews 75, 293306.
Santos, J.M., Sa, L.M., Reis Jr., N.C., Goncalves, R.F., Siqueira, R.N.,
2006. Modelling hydrogen sulphide emission in a WWTP with
UASB reactor followed by aerobic biofilters. Water Science and
Technology 54 (9), 173180.
Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and
Physics: From Air Pollution to Climate Change. WileyInterScience, USA.
Silva, H.P.M., 2003. Sulphur Transformations in the Liquid and
Gas Phases Inside a Aerated Biofilter Treating an UASB
Effluent (Transformacoes do enxofre nas fases lquida e
gasosa de um biofiltro aerado submerso tratando efluente de

um reator anaerobio do tipo UASB). M. Sc. thesis, Federal

University of Espirito Santo, Brazil.
USEPA, 1994. Air Emission Models for Waste and Wastewater.
North Carolina, USA. EPA-453/R-94080A. http://www.epa.
gov/ttn/chief/software/water/ (accessed 26.09.02.).
USEPA, 1995. Stationary Point and Area Sources. AP-42. In:
Compilation of Air Pollutant Emission Factor, fifth ed., vol. I.
Office of Air Quality Planning And Standards, North
Carolina, USA. http://www.epa.gov/ttn/chief/ap42/ (accessed
18.04.02.).
USEPA, 2001. Users Guide for Water9 Software. Version 2.0.0.
Office of Air Quality Planning and Standards, Research
Triangle Park, North Carolina. http://www.epa.gov/ttn/chief/
software/water/index.html (accessed 26.09.02.).
Van Langenhove, H., De heyder, B., 2001. Biotechnological
treatment of sewage odours. In: Stuetz, R., Frenchen, F.B. (Eds.),
Odours in Wastewater Treatment. IWA Publishing, UK, pp.
395412.
Wilmot, P.D., Cadee, K., Katinic, J.J., Kavanagh, B.V., 1988. Kinetics
sulfide oxidation by dissolved oxygen. Journal of Water
Pollution Control Federation 60 (7), 12641270.
Zhang, J.Z., Millero, F.J., 1993. The products from the oxidation of
H2S in seawater. Geochimica et Cosmochimica Acta 57 (8),
17051718.