ORIGINAL PAPER
Optimization of biological sulfide removal in a CSTR bioreactor
Aliakbar Roosta • Abdolhossein Jahanmiri •
Dariush Mowla • Ali Niazi • Hamidreza Sotoodeh
Received: 10 November 2011 / Accepted: 9 January 2012 / Published online: 18 January 2012
� Springer-Verlag 2012
Abstract In this study, biological sulfide removal from
natural gas in a continuous bioreactor is investigated for
estimation of the optimal operational parameters. According
to the carried out reactions, sulfide can be converted to ele-
mental sulfur, sulfate, thiosulfate, and polysulfide, of which
elemental sulfur is the desired product. A mathematical model
is developed and was used for investigation of the effect of
various parameters on elemental sulfur selectivity. The results
of the simulation show that elemental sulfur selectivity is a
function of dissolved oxygen, sulfide load, pH, and concen-
tration of bacteria. Optimal parameter values are calculated
for maximum elemental sulfur selectivity by using genetic
algorithm as an adaptive heuristic search. In the optimal
conditions, 87.76% of sulfide loaded to the bioreactor is
converted to elemental sulfur.
Keywords Sulfide removal � Bioreactor � Optimization �Genetic algorithms � Sulfur selectivity �Thiobacillus thioparus
List of symbols
A Constant of extended Debye–Huckel equation
(0.509 mol-� L�)
a Radius of the ion (m)
CB Optical density of bacteria
Fin Input flow rate (ml h-1)
I The ionic strength (mmol L-1)
k1 Reaction rate constant (mmol L-1 h-1)
k2 Reaction rate constant (mmol L-1)
k3 Reaction rate constant(mmol L-1)
k4 Reaction rate constant (mmol L-1 h-1)
k5 Reaction rate constant (mmol L-1)
k6 Reaction rate constant(mmol L-1)
k7 Reaction rate constant (mmol L-1)
k8 Reaction rate constant (mmol-0.59 L0.59 h-1)
pKx The acid dissociation constant
r Reaction rate (mmol L-1 h-1)
t Time (h)
V Volume of the broth in the bioreactor (L)
x Chain length of polysulfide ion
z The ionic charge of ion
b Constant of extended Debye–Huckel equation
(0.328 9 108 mol-1/2 L1/2 m-1)
c Activity coefficient
Introduction
Hydrogen sulfide is emitted by many industries, such as
petroleum refining, natural gas and petrochemical plants,
with very low odor-threshold value [1]. It is an extremely
toxic gas and has potential for injuring developing central
nervous systems at low-dose exposures [2]. The threshold
limit value for air 0.5–10 ppbv [3], natural gas 4 ppmv [4]
and for fresh or salty water fish is 0.5 ppm [5].
A. Roosta � A. Jahanmiri (&) � H. Sotoodeh
School of Chemical and Petroleum Engineering,
Shiraz University, Shiraz, Iran
e-mail: [email protected]
A. Roosta
e-mail: [email protected]
H. Sotoodeh
e-mail: [email protected]
D. Mowla
Environmental Research Center in Petroleum and Petrochemical
Industry, Shiraz University, Shiraz, Iran
e-mail: [email protected]
A. Niazi
Biotechnology Research Center, Shiraz University, Shiraz, Iran
e-mail: [email protected]
123
Bioprocess Biosyst Eng (2012) 35:1005–1010
DOI 10.1007/s00449-012-0685-5
The removal of hydrogen sulfide has been accomplished
using physical or chemical methods but, in the recent years
biological removal of hydrogen sulfide by microorganisms
at ambient temperatures and pressures is shown to be an
interesting alternative [6]. A review on bacteria of the
sulfur cycle was discussed by Tang et al. [7]. Also, a
review on removal of H2S from gas streams using bio-
logical processes was discussed by Sayed et al. [8]. In this
study, Thiobacillus thioparus (DSMZ 5368) was used (as
sulfur-oxidizing bacteria) for oxidation of hydrogen sulfide
into elemental sulfur. There are different types of biore-
actors, which could be used for biological sulfide removal,
the more common types are: bioscrubber, biotrickling fil-
ter, and biofilter. In the case of hydrogen sulfide removal
from natural gas, bioscrubber is more suitable than two
other processes. In a bioscrubber system (as shown in
Fig. 1), the sour gas is passed through a gas absorber where
H2S is washed from the gas stream by an alkaline such as
NaOH (Eqs. 1, 2), and then the rich alkaline solution is sent
to a bioreactor where the dissolved sulfide (HS-) is oxi-
dized to elemental sulfur or sulfate (Eqs. 3, 4) [9].
H2S gð Þ �! H2S aqð Þ ð1Þ
H2S aqð Þ þ OH� �! HS� þ H2O ð2Þ
HS� þ 1
2O2 �!
r1S0 þ OH� ð3Þ
S0 þ OH� þ 3
2O2 �!
r2SO2�
4 þ Hþ ð4Þ
In this system, in addition to the biological oxidation of
sulfide to sulfur and sulfate, undesirable abiotic reactions
occur in the bioreactor as shown in Eqs. 5, 6 [10]:
HS� þ ðx� 1ÞS0 $r3; r�3S2�
x þ Hþ ð5Þ
S2�x þ
3
2O2 �!
r4S2O2�
3 þ ðx� 2ÞS0 ð6Þ
According to these equations, dissolved sulfide can react
with produced S0 to producep polysulfide ions S2�x
� �; and
then S2�x ions are oxidized to S0 and S2O2�
3 .
The hydroxyl ions, consumed in the absorption of H2S,
are regenerated upon oxidation of sulfide to elemental
sulfur. This saves costs of dosing NaOH to the process. In
addition, elemental sulfur is easily separated from the
solution by sedimentation, and the produced elemental
sulfur can be used. When sulfate or thiosulfate is produced,
sodium hydroxide cannot be regenerated, and a stream of
hydroxide should be dosed to the bioreactor and a bleed
stream is necessary to prevent accumulation of sulfate and
thiosulfate. This leads to a considerable cost for the pro-
cess. Knowledge of the kinetics of the reactions helps to
prevent sulfate and thiosulfate formation. One of the most
important parameter which affects sulfate and thiosulfate
selectivity is the dissolved oxygen (DO) [11–13]. Ele-
mental sulfur selectivity increases by decrease of DO value
in the bioreactor. In addition, pH value, the amount of
sulfide load and concentration of bacteria have important
roles in sulfur selectivity.
The industrial sulfide removal process operates in
steady-state condition, and the operating parameters should
be at optimal values. Optimization as a major quantitative
tool is used in industrial decision making. In the present
study, the optimal condition of biological sulfide removal is
investigated. The goal of the optimization is the maximum
sulfur selectivity as shown in Eq. 7:
Sulfur selectivity ¼ ½S0�½HS��in
ð7Þ
Methods
A model was proposed and validated by Roosta et al. [14]
for biological sulfide removal in a fed-batch bioreactor
using the bacterium T. thioparus. The bacteria were pre-
served in a medium culture containing thiosulfate as an
energy source in a 5-day batch-inoculation period. After
inoculation with thiosulfate, the bacteria were adapted to
sulfide in a 36-h fed-batch period.
In their experiments, the bioreactor was operated in a
well-stirred condition, because a recirculating gas with a
flowrate of 15 L min-1 was spread by a diffuser; this
caused a good mixing of the broth. According to this fact,
they assumed that kinetic limitation is taken into account,
thus they ignored the interfacial mass transfer process of
gaseous oxygen as well as the dissolution of the solid
elemental sulfur. In addition, the density of broth was
assumed to be constant during the process. Also, theFig. 1 Schematic diagram of bioscrubber for biological sulfide
removal from natural
1006 Bioprocess Biosyst Eng (2012) 35:1005–1010
123
operating conditions of the bioreactor are listed in Table 1.
According to these assumptions and operational conditions,
the model proposed by Roosta et al. [14] is employed to
find the optimal steady-state operational conditions of the
process. The model equations are rewritten for steady-state
condition as follows:
½S2�x � ¼ 10�pKx
½HS��½Hþ�
cHS�
cHþcS2�x
ð8Þ
logci ¼ �ziA
ffiffiIp
1þ bai
ffiffiIp ð9Þ
I ¼ 1
2
Xz2
i ½i� ð10Þ
Finð½HS��in � ½HS��ÞV
� r1 � r4 ¼ 0 ð11Þ
�FinS0
Vþ r1 � r2 � r4 ¼ 0 ð12Þ
�Fin½SO2�4 �
Vþ r2 ¼ 0 ð13Þ
�Fin½S2O2�3 �
Vþ r4 ¼ 0 ð14Þ
r1 ¼CBk1½HS��k2 þ ½HS��
O2
k3 þ O2
ð15Þ
r2 ¼CBk4½S0�k5 þ ½S0�
O2
O2 þ k6
½OH��½OH�� þ k7
ð16Þ
r4 ¼ k8½S2�x �O0:59
2 ð17Þ
In this model, Eq. 5 assumed to be an equilibrium
reaction which the equilibrium is shown in Eqs. 8–10.
According to this assumption, by making material balances
on the present component in the bioreactor (HS-, S0,
SO42-, and S2O3
2-), Eqs. 11–14 are obtained for steady-
state condition and the reaction rates are given by Eqs. 15–
17. Gonz0alez-S0anchez and Revah [15] reported that the
inhibition of sulfide on microorganisms was detected at
sulfide concentration more than 3 mM; thus, the sulfide
inhibition is ignored in Eq. 15 due to low concentration of
sulfide in this study. Also, proposed models for sulfur and
sulfate production rates (Eqs. 15, 16) were validated for the
condition listed in Table 1; specially, pH dependence of
Eq. 16 is just true at pH values between 7.5 and 8.5, and the
elemental sulfur dependence of Eq. 16 is validated when the
particle sizes of elemental sulfur was less than 1 lm.
The model parameters obtained by Roosta et al. [14] are
listed in Table 2.
As mentioned before, the goal of the present study is the
maximum selectivity of elemental sulfur, which leads to
minimum hydroxide loss. In the following, the effect of
various parameters on sulfur selectivity is investigated to
find the optimum operational condition.
Results and discussion
Effect of sulfide load and DO on sulfur selectivity
According to Eqs. 3 and 4, sulfur is an intermediate in
oxidation of sulfide to sulfate. On the other words, these
equations are in series which the intermediate is the desired
reaction product. In this part, effects of sulfide load and DO
on sulfur selectivity are investigated. As illustrated in
Fig. 2, by increasing sulfide load, sulfur selectivity
increases, and passes a maximum, and then it decreases
with the increase of sulfide load. Furthermore, sulfur
selectivity decreases by increasing DO.
At low sulfide loads, the concentration of uneliminated
sulfide ([HS-]) is low as shown in Fig. 3; thus, rate of sulfur
production (r1), which is dependent on ([HS-] is low too.
However, rate of sulfate production (r2) is independent of
([HS-] and can be highly relative to r1; thus, large amount of
elemental sulfur is converted to sulfate. Consequently,
Table 1 Bioreactor operating
conditionsTemperature (�C) OD600 of bacteria DO (ppm) pH
30 ± 0.5 0.3–0.6 0.4–6 8 ± 0.5
Sulfide load (mmol L-1 h-1) Uneliminated sulfide (mM) Flowrate (ml h-1)
0.3–5.7 up to 1 1.5–23
Table 2 Parameters of model equations [14]
k1 (mmol L-1 h-1) k2 (mmol L-1) k3 (mmol L-1) k4 (mmol L-1 h-1) k5 (mmol L-1)
10.051 0.106 2.796 9 10-5 16.091 2.524
k6 (mmol L-1) k7 (mmol L-1) k8 (mmol-0.59 L0.591 h-1) pKx
0.203 5.593 9 10-4 39.751 8.96
Bioprocess Biosyst Eng (2012) 35:1005–1010 1007
123
sulfate selectivity is high at low sulfide loads as shown in
Fig. 4. Increase of [HS-] by increasing sulfide load leads to
increase of r1 and sulfur selectivity passes a maximum.
As the sulfide load is increased, the concentration of
polysulfide ions [Sx2-] increase according to Eq. 8 and it
leads to increase of thiosulfate production as shown in
Fig. 5. As thiosulfate production consumes elemental sul-
fur, increase of thiosulfate selectivity is the main reason for
the decrease of sulfur selectivity at higher sulfide loads.
The effect of DO on sulfur, sulfate, and thiosulfate
selectivity is shown in Figs. 2, 4 and 5.
In addition, by increasing DO value, the maximum
sulfur selectivity occurs at higher sulfide load as shown by
filled circles (•) in Fig. 2.
Effect of pH on sulfur selectivity
As shown in Fig. 6, by increasing pH value, sulfur selec-
tivity decreases, which is due to two mechanisms. Firstly,
increasing [OH-] leads to increase r2 according to Eq. 4;
thus, more elemental sulfur is oxidized to sulfate. Sec-
ondly, at high pH values, thiosulfate selectivity increases
(the concentration of thiosulfate increases due to increase
of polysulfide concentration), and more parts of produced
sulfur are converted to thiosulfate. The effect of pH on
thiosulfate selectivity is illustrated in Fig. 7.
Effect of bacteria OD on sulfur selectivity
Sulfur selectivity as a function of sulfide load at different
bacteria ODs is investigated and is shown in Fig. 8. For
low sulfide loads (below 2.4 mmol L-1 h-1), sulfur
selectivity decreases by increasing bacteria concentration
at a known sulfide load. At sulfide load about
2.4 mmol L-1 h-1, sulfur selectivity is almost independent
of bacteria concentration, and there is a direct relation
between sulfur selectivity and bacteria OD at higher sulfide
loads. At a known concentration of bacteria, sulfur
Fig. 2 Ratio of [S0] to [HS-]in as a function of [HS-] load at various
DO values. Filled circles represents peak of curves
Fig. 3 Uneliminated HS- concentration as a function of HS- load at
various DO values
Fig. 4 Ratio of [SO42-] to [HS-]in as a function of [HS-] load at
various DO values
Fig. 5 Ratio of [S2O32-] to [HS-]in as a function of [HS-] load at
various DO values
1008 Bioprocess Biosyst Eng (2012) 35:1005–1010
123
selectivity increases with the increase of sulfide load, and
it passes through a maximum, and then it decreases with
the increase of sulfide load as discussed before. The
maximum sulfur selectivity occurs at higher sulfide load
for higher bacteria OD as shown by filled circles (•) in
Fig. 8.
Optimization of process using genetic algorithm (GA)
According to the above sections, various parameters can
affect sulfur selectivity such as: pH, DO value, OD of
bacteria, and sulfide load. In this part of study, the optimum
conditions of the bioreactor are estimated using genetic
algorithm optimization method. Genetic algorithms are
efficient in getting good solutions for difficult non-linear
optimization problems [16]. They are search methods that
simulate natural evolution patterns [17]. A simple GA was
first expressed by Holland [18] and further developed by
Goldberg [19]. Since then, several amendments were made
to turn the initial idea of genetic algorithms to more effi-
cient optimization algorithms, very useful information can
be found in Michalewicz [20]. In the recent years, GA
method has been applied to optimization of many of bio-
technology and biochemical engineering processes
[21–25].
As mentioned previously, the objective of the optimi-
zation is the maximum sulfur selectivity, which should be
maximized and the decision variables are considered as OD
of bacteria, DO value, sulfide load, and pH value. The
choices of GA parameters have a great effect on the speed
of convergence and success of the optimization (finding a
global optimum instead of local optimum). The exact set-
tings used for running the GA (number of individuals in the
population, mutation rate, migration fraction, and cross-
over fraction) are listed in Table 3. The best objective
function was achieved at less time with using these par-
ticular parameters.
The obtained optimal values of decision variables are
shown in Table 4, and the selectivity of the products at
optimal conditions are shown in Table 5. Sulfur selectivity
is in the maximum value (87.76%) in the optimal
condition.
Fig. 6 Effect of pH on sulfur selectivity at various DO values
Fig. 7 Effect of pH on thiosulfate selectivity at various DO values
Fig. 8 Effect of bacteria OD on sulfur selectivity, pH = 8,
DO = 0.8 ppm. Filled circles represents peak of curves
Table 3 Parameters of GAPopulation size Mutation rate Cross over Migration No. of generations
Function fraction Direction fraction
30 0.03 Scattered 0.75 Forward 0.2 1,000
Bioprocess Biosyst Eng (2012) 35:1005–1010 1009
123
Conclusion
In the present study, optimization of biological sulfide
removal from natural gas is successfully performed using
GA. The objective of the optimization is sulfur selectivity
and the decision variables are bacteria OD, dissolved
oxygen, pH, and sulfide load. Optimization results show
that at optimal condition, maximum 87.76% of sulfide
loaded to the bioreactor could be converted to elemental
sulfur and the remained is converted to sulfate and thio-
sulfate. At this condition, for each 100 mol of sulfide load,
12.24 mol hydroxide ions are spent due to sulfate and
thiosulfate production.
References
1. Aroca G, Urrutia H, Nunez D, Oyarzun P, Arancibia A, Guerrero
K (2007) Comparison on the removal of hydrogen sulfide in
biotrickling filters inoculated with Thiobacillus thioparus and
Acidithiobacillus thiooxidans. J Biotech 10:514–520
2. Roth SH, Skrajny B, Reiffenstein RJ (1995) Alteration of the
morphology and neurochemistry of the developing mammalian
nervous system by hydrogen sulphide. Clin Exp Pharmacol
22:379–380
3. Ruth JH (1986) Odor thresholds and irritation levels of several
chemical substances: a review. Am Ind Hyg Assoc J 47:142–151
4. Kohl AL, Nielsen RB (1997) Gas purification. Gulf Publishing
Company, Texas
5. Tian S, Zhang L, Wang Q, Wang X, Xie W (2007) Study on
hydrogen sulfide removal based on bench-scale experiment by
bio-trickling filter. J Zhejiang Univ Sci A 8:1141–1148
6. Bouzaza A, Laplanche A, Marsteau S (2004) Adsorption oxida-
tion of hydrogen sulfide on activated carbon fibers: effect of the
composition and the relative humidity of the gas phase. Che-
mosphere 54:481–488
7. Tang K, Baskaran V, Nemati M (2009) Bacteria of the sulphur
cycle: an overview of microbiology biokinetics and their role in
petroleum and mining industries. Biochem Eng J 44:73–94
8. Sayed M, Soreanu G, Falleta P, Beland M (2006) Removal of
hydrogen sulfide from gas streams using biological processes: a
review. Can Biosyst Eng 48:2–14
9. Gadre RV (1989) Removal of hydrogen sulfide from biogas by
chemoautotrophic fixed-film bioreactor. Biotechnol Bioeng
34:410–414
10. Kleinjan WE, de Keizer A, Janssen AJH (2005) Kinetics of the
reaction between dissolved sodium sulfide and biologically pro-
duced sulfur. Ind Eng Chem Res 44:309–317
11. van den Bosch PLF, van Beusekom OC, Buisman CJN, Janssen
AJH (2007) Sulfide oxidation at halo-alkaline conditions in a fed-
batch bioreactor. Biotechnol Bioeng 97:1053–1063
12. Kleinjan WE, de Keizer A, Janssen AJH (2003) Biologically
produced sulfur. Top Curr Chem 230:167–188
13. Buisman CNJ, Gerrats BG, Ijspeert P, Lettinga G (1990) Opti-
mization of sulphur production in a biotechnological sulphide
removing reactor. Biotechnol Bioeng 35:50–56
14. Roosta A, Jahanmiri A, Mowla D, Niazi A (2011) Mathematical
modeling of biological sulfide removal in a fed batch bioreactor.
Biochem Eng J 58–59:50–56
15. Gonz0alez-S0anchez A, Revah S (2007) The effect of chemical
oxidation on the biological sulfide oxidation by an alkaliphilic
sulfoxidizing bacterial consortium. Enzyme Microb Tech
40:292–298
16. Park C, Lee TY (2004) Optimal control by evolutionary algo-
rithm technique combined with spline approximation method.
Chem Eng Commun 91:262–277
17. Rezaei N, Kordabadi H, Elkamel A, Jahanmiri A (2009) An
optimal extended Kalman filter designed by genetic algorithms.
Chem Eng Comm 196:602–615
18. Holland JH (1975) Adaptation in natural and artificial systems.
University of Michigan Press, Michigan
19. Goldberg DE (1989) Genetic algorithms in search optimization
and machine learning. Addison-Wesley Reading, Boston
20. Michalewicz Z (1996) Genetic algorithms data structures evolu-
tion programs. Springer, New York
21. Rao CS, Sathish T, Mahalaxmi M, Laxmi GS, Rao SR, Prak-
asham RS (2007) Modeling and optimization of fermentation
factors for enhancement of alkaline protease production by iso-
lated Bacillus circulans using feed-forward neural network and
genetic algorithm. J Appl Microbiol 104:889–898
22. Na JG, Chang Y, Chung B, Lim H (2001) Adaptive optimization
of fed-batch culture of yeast by using genetic algorithms. Bio-
process Biosyst Eng 24:299–308
23. Lee FC, Rangaiah GP, Ray AK (2007) Multi-objective optimi-
zation of an industrial penicillin V bioreactor train using non-
dominated sorting genetic algorithm. Biotechnol Bioeng
98:586–598
24. Etschmann MMW, Sell D, Schrader J (2004) Medium optimi-
zation for the production of the aroma compound 2-phenylethanol
using a genetic algorithm. J Mol Catal B Enzym 29:187–193
25. Liu G, Zhou D, Xu H, Mei C (2010) Model optimization of SVM
for a fermentation soft sensor. Expert Syst Appl 37:2708–2713
Table 4 Optimal values of the process variables
OD of bacteria DO (ppm) Sulfide load (mmol L-1 h-1) pH
0.39 0.40 2.22 7.80
Table 5 Selectivity of products at optimal conditions
Sulfur particles Sulfate Thiosulfate Uneliminated sulfide
87.76% 8.53% 3.65% 0.06%
1010 Bioprocess Biosyst Eng (2012) 35:1005–1010
123