See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330511023

Kinetic study of the regeneration of spent caustic via the genetic algorithm
method

Article in Environmental Health Engineering and Management · November 2018

DOI: 10.15171/EHEM.2018.31

CITATIONS READS

2 238

4 authors:

Asadollah Karimi Esmaeil Fatehifar

University of Maragheh Sahand University of Technology
16 PUBLICATIONS 124 CITATIONS 97 PUBLICATIONS 1,281 CITATIONS

SEE PROFILE SEE PROFILE

Reza Alizadeh Hadi Soltani

Shiraz University Islamic Azad University, Ahar Branch, Iran, Ahar
25 PUBLICATIONS 247 CITATIONS 24 PUBLICATIONS 297 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hadi Soltani on 16 July 2019.

The user has requested enhancement of the downloaded file.

Environmental Health Engineering and Management Journal 2018, 5(4), 231–239 doi 10.15171/EHEM.2018.31
http://ehemj.com

Environmental Health
HE
Engineering and
MJ
Management Journal Original Article
Open Access
Publish Free

Kinetic study of the regeneration of spent caustic via the genetic

algorithm method
ID ID
Asadollah Karimi1 , Esmaeil Fatehifar2* , Reza Alizadeh2, Hadi Soltani3
1
Department of Chemical Engineering, Faculty of Engineering, University of Maragheh, Maragheh, Iran
2
Environmental Engineering Research Center, Faculty of Chemical Engineering, Sahand University of Technology, Tabriz, Iran
3
Department of Chemical Engineering, Ahar Branch, Islamic Azad University, Ahar, Iran

Abstract Article History:

Background: Spent caustic contains noxious components such as sulfide species and also high chemical Received: 21 August 2018
oxygen demand content (COD). Oxidation of these materials to caustic and sulfate species is mostly the Accepted: 14 October 2018
rate-controlling step within catalytic oxidation of spent caustic. ePublished: 7 November 2018
Methods: In this study, the kinetics of catalytic oxidation of spent caustic and the regeneration methodology
of the sulfidic spent caustic were investigated. The kinetics of catalytic oxidation of spent caustic was studied
in the presence of a heterogeneous catalyst. The developed mathematical model was verified via the batch
bubble column reactor. The elementary and non-elementary models based on the genetic algorithm were
used to obtain the rate coefficient and kinetic order.
Results: The experiments were carried out at various conditions. The results indicated that the error of
objective function of the non-elementary and elementary models was 3.01% and 134.96%, respectively.
Conclusion: According to the results, the non-elementary model had rational outcome compared to the
elementary one. Also, non-elemental model is more concordance with experimental results.
Keywords: Caustic, Kinetic, Regeneration, Catalysis
Citation: Karimi A, Fatehifar E, Alizadeh R, Soltani H. Kinetic study of the regeneration of spent caustic
via the genetic algorithm method. Environmental Health Engineering and Management Journal 2018; 5(4): *Correspondence to:
231–239. doi: 10.15171/EHEM.2018.31. Esmaeil Fatehifar
Email: [email protected]

Introduction oxidation (WAO), advanced oxidation process (AOP)

Chemical industries are among those fields which create and catalytic oxidation process (COP) that have been
various kinds of organic and inorganic waste in varying applied to treat spent caustic wastewater (6-14). WAO is
concentrations that must be treated. The contamination carried out at high temperature and pressure depending
of these wastes is not only limited to water, air, soil and on feed. Often, WAO is applied for pre-treating of spent
noise, but also has extended to the human’s life (1,2). caustic and it cannot lower the chemical oxygen demand
The petrochemical plants are among those industries (COD) of spent caustic to desired effluent (15). Therefore,
that generate all kinds of wastewater. The aqueous it is necessary to use catalytic wet air oxidation (CWAO).
caustic solutions are usually applied for the removal of Treatment of spent caustic wastewater was studied in
impurities, like sulfur compounds and carbon dioxide catalytic reactor with graphene oxide (GO) and ruthenium/
in petrochemical plants such as the olefin unit (3). The GO catalysts (16). AOP can be defined as the acceleration
effluent of caustic scrubbing as a waste solution is known as of an oxidation reaction by generation of hydroxyl radicals.
spent caustic. Spent caustic is a highly deadly and odorous Also, AOP are defined as processes which treat wastewater
waste. According to the US Environmental Protection steams with high COD and toxic contaminants (17). An
Agency (EPA) reports, spent caustic is categorized as a AOP system involves a catalyst (photo-catalysis) and an
hazardous and deleterious waste in the olefin unit (4). oxidant such as H2O2, O3/H2O2, fenton, and UV. A novel
Table 1 shows some characteristics of spent caustic from photocatalytic reactor was studied for the treatment of
the olefin unit in a petrochemical plant. spent caustic wastewater and TiO2-clinoptilolite was
There are a couple of methods for treating spent caustic, synthesized for the enhancement of its photocatalytic
such as incineration, disposal and oxidation methods, efficiency (18). COP is used to remove sulfide compounds
among which chemical oxidation is very practical and from the spent caustic wastewater. In the previous study,
applicable (5). Some prominent methods are wet air spent caustic wastewater of an olefin unit was treated in

© 2018 The Author(s). Published by Kerman University of Medical Sciences. This is an open-access article distributed under
the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
Karimi et al

Table 1. Characteristics of spent caustic produced by Olefin unit

Item Value
Total sulfur 14000-21000 mg/L
COD 15000-30000 mg/L
pH 13.5-13.7
S2- 670 mg/L
Phenols 300 mg/L
Specific gravity 1.1
NaOH 1-3 wt%

two reactor (the system can be represented via a series

of bubble column reactor with mixed reactor) by the
catalytic chemical oxidation method (19). This waste Figure 1. The flowchart of the overall kinetics.
stream was finally regenerated and recycled to the unit.
In the present study, the global reaction kinetics were (Olefin unit) in Iran. The catalytic chemical oxidation was
modelled and optimized in a bubble column reactor via conducted to regenerate and treat spent caustic in a bubble
genetic algorithm (GA). Analysis of the kinetic models column reactor, with capacity of 500 ml and 40 µm pore
of catalytic reactions requires more accurate estimation size (Figure 2). IVKAZ catalyst of the JSC company was
of rate parameters and constants (20-22). The most employed to regenerate spent caustic in a bubble column
commonly used methods in determining the optimal reactor. The reactor was operated at ambient pressure and
parameters model, are non-linear fitting algorithm like 90°C. In previous work, the treatment of spent caustic
artificial neural networks and Marquardt algorithm. But, was accomplished in two reactors by the full factorial
the structure of non-linear reaction kinetics treatment of design (19). It has been found that the optimal operating
spent caustic has more than one relative minimum value. temperature and air flow for treatment and regeneration of
So, it is difficult to achieve the optimal point by using spent caustic were 85 to 95°C and 10 L/min, respectively,
these methods. One way to solve such kind of problems while temperature of 90°C and air flow of 10 L/min were
is the use of randomized algorithms. As a new method, chosen for this kinetic study.
GA has been introduced for optimization of non-linear All the demanded materials purchased from Merck
models (23-25). Various applications of GA have been Chemicals were used for data analysis procedure. The
proposed in non-linear models to determine the optimal mercaptide and hydrosulfite sodium were measured using
parameters. The flexibility of genetic algorithm is its a digital pH meter (Metrohm). These compounds were
major advantage which provides the objective function determined according to ASTM-UOP 209. Thiosulfate
and constraints in farming (24). It can be seen that was measured using a spectrophotometer. The standard
researchers have used GA to estimate the amounts of thiosulfate titrant and standard iodine titrant, HCl and
kinetic parameters of complex reactions. Finding a general indicator were used for the titration of S2- and alkalinity,
kinetic model for hydrogenolysis of dibenzothiophene respectively. The alkalinity, S2- and O2 of the spent caustic
(DBT) based on Langmuir-Hinshelwood type (26), and were analyzed by standard methods for the examination
estimating simultaneously the kinetic as well as energetic of water and wastewater (28). All solutions were prepared
parameters of the complex reaction of the saponification using deionized water.
of ethyl acetate (27) are examples of the use of genetic
algorithm in these cases. In this study, a bench-scale pilot
was employed for regenerating and treating spent caustic.
The development of global reaction of spent caustic is the
main purpose of this work. Another purpose of this study
was to investigate the application of GA to choose a precise
kinetic model and determine the optimal parameters.
The parameters of this reaction are rate constants and
kinetic order of concentrations. The kinetics of the
reaction was surveyed as a global reaction in a bubble
column reactor. This function objective is prediction error
of the model and the experimental concentration. The
kinetic steps are shown in Figure 1.

Materials and Methods

Experimental procedure
The waste solution was provided by a petrochemical plant Figure 2. The Schematic of bench scale pilot.

232 Environmental Health Engineering and Management Journal 2018, 5(4), 231–239
Karimi et al

Kinetic modelling In this study, the following steps were performed to

Spent caustic contains a wide variety of compounds, each determine the kinetic parameters. At first, members of the
one has its own specific kinetic characteristics. In the primary population including constants of reactions and
majority of the reactions (1 to 5), thiosulfate was produced kinetic order of reaction were generated in a completely
as an intermediate compound. Several previous studies random manner, depending on the range of defined
have shown that thiosulfate converts to sulfate in an changes. This vector consists of seventeen variables
alkaline environment at 76-85°C. The reaction rates 1 to and five variables for non-elementary reactions and
5 was considered as both elementary and non-elementary elementary reactions, respectively. An example of these
reactions. The GA method was used for each of the vectors is Vi = (k1 k2 k3 k4 k5 X6 X7 X8 X9 X10 X11 X12 X13 X14
reactions. Table 2 indicated the reaction rate equations. X15 X16 X17) for non-elementary reaction and Vi = (k’1 k’2
The reaction rate equations of a, b, c, d, and f belong to k’3 k’4 k’5) for elementary reaction. The range of variables
caustic, thiosulfate sodium, oxygen, hydrosulfite sodium, changes inside the vectors were between (10-3 - 10+1) and
mercaptide, and sodium sulfide, respectively. population size was 500.
As air flow enters during the process, the reactor is a semi The objective function is defined as the minimum error
batch. It is assumed ideal for studying the reaction kinetic model results and the experimental results. The error of
(29,30). The bubble column reactor was considered as a objective function is shown in equation 8.
batch reactor. For this purpose, the oxygen injection was n C exp − C cal
computed every 5 seconds and the total amount has been Error = 1 ∑ i i
*100 (8)
added in 20, 25, and 30 minutes. n i =1 Ciexp
In – Out + ri*V= d(Ci*V)/dt (6) After generating the initial populations via the GA,
Where ri is the rate of reaction for i material, V is the the value of the objective function of each member is
volume of the batch reactor and t is time. calculated. By using the objective function values based
Oxygen is the only component which enters the reactor on the roulette wheel method, members are ranked.
and other ones have no entery or exit flow. The solution Obviously, any member with a better objective function
volume (V) is not constant due to high evaporation will be considered more valuable. GA operators are
especially at high temperatures. To solve this problem, reproduction, crossover and mutation (31,32).
the reaction time is divided to shorter intervals (some Reproduction: By this operator, 50% of the members of
seconds) and the output of each interval assumes as the the current generation based on the roulette wheel are
input of the next interval. This technique removes the transferred to the new generation as follows:
input and output terms from equation 6 as mentioned I) Based on elitism: The best members of the current
below: generation (5%) are transferred to the new generation
1- At the beginning of each interval, it is assumed that the directly (and not randomly),
total amount of oxygen which was supposed to enter the II) 45% of the members are randomly assigned to the new
reactor gradually, is injected at the beginning. generation (30).
2- As the reaction time is divided to shorter intervals, the Crossover: To produce new generation by the crossover
volume variations at these intervals will be small enough operation, both one- and two-point crossovers are
to be neglected. It is worth noting that the shorter intervals employed. After selecting two vectors by the roulette
to be considered, the error of this method will be less. In wheel, a random number is generated to decompose
this way, the equation 7 is rearranged as follows: the chromosome (i.e. the vector) into several pseudo
ri = d(Ci)/dt (7) parents. Then, these parents are composed together to

Table 2. The equation rate of the reactions (1 to 5)

Non-elementary Elementary
𝑋𝑋 𝑋𝑋 𝑋𝑋 𝑋𝑋 𝑋𝑋
1- 𝑋𝑋 𝑋𝑋
𝑅𝑅𝑎𝑎 = −𝑘𝑘1 𝐶𝐶𝑎𝑎 6 𝐶𝐶𝑏𝑏 7 𝐶𝐶𝑐𝑐 8 − 𝑘𝑘2 𝐶𝐶𝑎𝑎 9 𝐶𝐶𝑐𝑐 10 𝐶𝐶𝑑𝑑 11 + 2𝑘𝑘4 𝐶𝐶𝑒𝑒 14 𝐶𝐶𝑐𝑐 15 +
𝑋𝑋
1- 𝑅𝑅𝑎𝑎 = −𝑘𝑘 1́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑐𝑐 𝐶𝐶𝑏𝑏0.5 − 𝑘𝑘 2́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑐𝑐2 𝐶𝐶𝑑𝑑 + 2𝑘𝑘 4́ 𝐶𝐶𝑒𝑒2 𝐶𝐶𝑐𝑐0.5 +
𝑋𝑋 𝑋𝑋
2𝑘𝑘5 𝐶𝐶𝑐𝑐 16 𝐶𝐶𝑓𝑓 17 2𝑘𝑘 5́ 𝐶𝐶𝑐𝑐 𝐶𝐶𝑓𝑓
𝑋𝑋 𝑋𝑋
2- 𝑋𝑋
𝑅𝑅𝑏𝑏 = −0.5𝑘𝑘1 𝐶𝐶𝑎𝑎 6 𝐶𝐶𝑏𝑏 7 𝐶𝐶𝑐𝑐 8 + 0.5𝑘𝑘3 𝐶𝐶𝑑𝑑 12 𝐶𝐶𝑐𝑐 13 𝐶𝐶𝑑𝑑 11 +
𝑋𝑋 𝑋𝑋 𝑋𝑋
2- 𝑅𝑅𝑏𝑏 = −0.5𝑘𝑘 1́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑏𝑏0.5 𝐶𝐶𝑐𝑐 + 0.5𝑘𝑘 3́ 𝐶𝐶𝑐𝑐2 𝐶𝐶𝑑𝑑 + 𝑘𝑘 5́ 𝐶𝐶𝑐𝑐 𝐶𝐶𝑓𝑓
𝑋𝑋 𝑋𝑋
𝑘𝑘5 𝐶𝐶𝑓𝑓 17 𝐶𝐶𝑐𝑐 16
𝑋𝑋 𝑋𝑋 𝑋𝑋 𝑋𝑋
3- 𝑋𝑋 𝑋𝑋
𝑅𝑅𝑐𝑐 = −𝑘𝑘1 𝐶𝐶𝑎𝑎 6 𝐶𝐶𝑏𝑏 7 𝐶𝐶𝑐𝑐 8 − 2𝑘𝑘2 𝐶𝐶𝑎𝑎 9 𝐶𝐶𝑐𝑐 10 𝐶𝐶𝑑𝑑 11 − 2𝑘𝑘3 𝐶𝐶𝑑𝑑 12 𝐶𝐶𝑐𝑐13 −
𝑋𝑋
3- 𝑅𝑅𝑐𝑐 = −𝑘𝑘 1́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑏𝑏0.5 𝐶𝐶𝑐𝑐 − 2𝑘𝑘 2́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑐𝑐2 𝐶𝐶𝑑𝑑 − 0.5𝑘𝑘 3́ 𝐶𝐶𝑑𝑑 𝐶𝐶𝑐𝑐2 −
𝑋𝑋 𝑋𝑋 𝑋𝑋 𝑋𝑋
0.5𝑘𝑘4 𝐶𝐶𝑒𝑒 14 𝐶𝐶𝑐𝑐 15 − 2𝑘𝑘5 𝐶𝐶𝑐𝑐 16 𝐶𝐶𝑓𝑓 17 0.5𝑘𝑘 4́ 𝐶𝐶𝑒𝑒2 𝐶𝐶𝑐𝑐0.5 − 2𝑘𝑘 5́ 𝐶𝐶𝑐𝑐 𝐶𝐶𝑓𝑓
𝑋𝑋 𝑋𝑋
4- 𝑋𝑋
𝑅𝑅𝑑𝑑 = −𝑘𝑘2 𝐶𝐶𝑎𝑎 9 𝐶𝐶𝑐𝑐 10 𝐶𝐶𝑑𝑑 11 − 𝑘𝑘3 𝐶𝐶𝑑𝑑 12 𝐶𝐶𝑐𝑐13
𝑋𝑋
4- 𝑅𝑅𝑑𝑑 = −𝑘𝑘 2́ 𝐶𝐶𝑎𝑎 𝐶𝐶𝑐𝑐2 𝐶𝐶𝑑𝑑 − 𝑘𝑘 3́ 𝐶𝐶𝑑𝑑 𝐶𝐶𝑐𝑐2

𝑋𝑋
5- 𝑋𝑋
𝑅𝑅𝑒𝑒 = −2𝑘𝑘4 𝐶𝐶𝑒𝑒 14 𝐶𝐶𝑐𝑐 15 5- 𝑅𝑅𝑒𝑒 = −2𝑘𝑘 4́ 𝐶𝐶𝑒𝑒2 𝐶𝐶𝑐𝑐0.5
𝑋𝑋
6- 𝑅𝑅𝑓𝑓 =
𝑋𝑋
−2𝑘𝑘5 𝐶𝐶𝑐𝑐 16 𝐶𝐶𝑓𝑓 17 6- 𝑅𝑅𝑓𝑓 = −2𝑘𝑘 5́ 𝐶𝐶𝑐𝑐 𝐶𝐶𝑓𝑓

Environmental Health Engineering and Management Journal 2018, 5(4), 231–239 233
Karimi et al

produce offspring. In this study, the rate of crossover was method is shown in Figure 3.
considered to be 50%.
It is noteworthy that in this operator, it is possible to Results
choose more members with a better objective function. The concentrations of components in the bubble
Mutation: New generation of the present operation is column reactor were obtained. As mentioned before, the
used to all members after producing. The definition temperature and flow rate of reaction were constant and
of the mutation and its rate plays an important role only the reaction time changed. The experimental amount
in the convergence of the algorithm. As a general of each component is indicated in Figure 4-9. The value
fact, inappropriate definition of this operator leads to of spent caustic is illustrated for elementary and non-
premature convergence of the GA. Therefore, mutation elementary models in Figure 4. At an earlier time (20
should be defined in such a manner that the GA has minutes), the elementary model can better predict the
sufficient diversity during its evolutionary procedure. The experimental results, also non-elementary model can be
mutation rate is based on the following equation: seen to be better with time.
Mu=Mumin+ (Mumax-Mumin). exp (-10*(Cw-Cb)/Cw) (9) The obtained results of thiosulfate sodium are
Where Mu is the mutation rate, calculated in each demonstrated in Figure 5. As shown in Figure 5, the
generation. Mumin and Mumax are the minimum and thiosulfate amount will be reduced by the increase of
maximum mutation rates allowed and their values are 10% reaction time. The oxidation process of thiosulfate to
and 100%, respectively, in the equation 9. Cw is the value of sulfate is a heterogeneous gas-liquid reaction.
the cost of the worst individual and Cb is the value of the Figure 6 compares oxygen amount botained by
cost of the best one (28). The flowchart of the applied GA experimental and GA methods. The real amount of

GA

Define Parameter
of GA

Generation of Initial
population

Apply Apply Apply Calculation of Fitness

Mutation Crossover Reproduction (Function Error)

No

Calculation of Fitness
Max Number
for New Member
Iteration Reached

Yes

End

Figure 3. The diagram of genetic algorithm.

Figure 3. The diagram of genetic algorithm.

Experimental Elementary Model Non-Elementary Model

0.042
Error of Non- Elem. Model= 14.8%
Error of Elem. Model= 10.4%
Molar Concenteration

0.039

0.036

0.033

0.03
18 20 22 24 26 28 30 32
Time (min)

18
Figure 4. The results of caustic obtained from experimental data and kinetic model.

234 Environmental Health Engineering and Management Journal 2018, 5(4), 231–239
Karimi et al

oxygen has been computed based on the stoichiometric experimentally-obtained results of NaHS. The hydrosulfite
calculations in a 500 mL bubble column reactor. As shown sodium is consumed in two different reactions, so the
in Figure 6, the results of the non-elementary model is prediction of this component is difficult by the elementary
near to the actual values. model.
According to the amount of NaHS shown in Figure 7, The oxidation of NaSR is illustrated in Figure 8. The non-
the elementary model is not a desirable model to predict elementary model has really low error compared to the

Experimental Elementary Model Non-Elementary Model

0.48
Error of Non- Elem. Model= 5.9%
0.42 Error of Elem. Model= 498%
Molar Concentration

0.36

0.3

0.24

0.18

0.12

0.06

0
18 20 22 24 26 28 30 32
Time (min)

Figure 5. The results of thiosulfate obtained from experimental data and kinetic model.

Experimental Elementary Model Non-Elementary Model

5.8
Error of Non- Elem. Model= 1.5%
5.3 Error of Elem. Model= 30%
Molar Concentration

4.8

4.3

3.8

3.3

2.8

2.3
18 20 22 24 26 28 30 32
Time (min)

Figure 6. The results of oxygen obtained from experimental data and kinetic model.

Experimental Elementary Model Non-Elementary Model

0.8
Error of Non- Elem. Model= 19%
0.7
Error of Elem. Model= 100%
Molar Conentration

0.6

0.5

0.4

0.3

0.2

0.1

0
18 20 22 24 26 28 30 32
Time (min)

Figure 7. The results of NaHS obtained from experimental data and kinetic model.

Environmental Health Engineering and Management Journal 2018, 5(4), 231–239 235
Karimi et al

Experimental Elementary Model Non-Elementary Model

0.02
Error of Non- Elem. Model= 9%
Error of Elem. Model= 94.6%

Molar Concentration
0.015

0.01

0.005

0
18 20 22 24 26 28 30 32
Time (min)

Figure 8. The results of NaSR obtained from experimental data and kinetic model.

elementary one. The results illustrate that non-elementary NaSH + 2O2 + NaOH Na2SO4 + H2O (2)
model is more concordance with experimental results NaSH + 2O2 ½ Na2S2O3 + ½ H2O (3)
(error of non-elementary model was lower than 10%). 2RSNa + ½ O2 + H2O RSSR + 2NaOH (4)
The change of amount of Na2S is indicated in Figure 9. Na2S + O2 + ½H2O NaOH + ½ Na2S2O3 (5)
The catalytic oxidation with increase of the reaction
time, led to higher conversion of sulfides to thiosulfate, One of the aims of this study was to investigate the
therefore, the amount of sulfide species reduced. The regeneration of spent caustic, therefore, the amount of
increase of reaction time helped approach the value of caustic was expected to increase. The present claim was
non-elementary model to the experimental amount and confirmed by the experimental data. The reaction time
resulted in a low error at the end of the reaction time. variables had a significant effect on the regeneration
The kinetic coefficients obtained from elementary and process of spent caustic. It is mostly proffered to increase
non-elementary models are shown in Table 3. The kinetic reaction time of the whole process. The error of the
coefficient of the reaction 4 had the highest value among elementary model was less than the non-elementary one,
other reactions. This value made the reaction rate much but the concentration of spent caustic was decreased. It
faster than the other reactions. is obvious that the elementary model couldn’t properly
Table 4 indicates the order of concentration in the non- predict the components of spent caustic in Figure 4.
elementary model. These values were obtained by the GA So, the experiments should be executed in a regime where
method. diffusional resistance in gas and liquid phases could be
ignored to find out the true kinetics (34). The kinetics
Discussion of catalytic oxidation was carried out in the presence of
Different chemical oxidation reactions took place in a cobalt phthalocyanine (IVKAZ) catalyst. As shown in
bubble column reactor. The main oxidation reactions of Figure 5, the thiosulfate content was significantly reduced,
spent caustic are as follows (33): so that the experimental data as well as the results of
elementary and non-elementary models’ showed the same
NaOH + ½ Na2S2O3 + O2 Na2SO4 + ½ H2O (1) trend regarding to the studied kinetics. Figure 5 showed

Experimental Elementary Model Non-Elementary

0.012 Error of Non- Elem. Model= 3.3%

Error of Elem. Model= 81.5%
0.01
Molar Concentration

0.008

0.006

0.004

0.002

0
18 20 22 24 26 28 30 32
Time (min)

Figure 9. The results of Na2S obtained from experimental data and kinetic model.

236 Environmental Health Engineering and Management Journal 2018, 5(4), 231–239
Karimi et al

Table 3. Kinetic rate coefficient value for the reactions (1-5) As shown in Table 3, the orders of the non-elementary
Kinetic rate Elementary model are quite different from the elementary model.
Non-elementary model
coefficient model The non-elementary model seems to display really better
k1 & k'1 k'1=0.0001 k1=0.0636 results compared with the elementary one. The overall
k2 & k'2 k'2=0.5658 k2=0.0313 error of the non-elementary model was less than the
k3 & k'3 k'3=0.4999 k3=0.0005 elementary one. The comparison of others shows that the
k4 & k'4 k'4=0.5000 k4=2.0000 order of concentration of the reaction (4) was reasonable.
k5 & k'5 k'5=0.1000 k5=0.3000 One of the reasons that spent caustic is regenerated
in these conditions is due to the value obtained in the
reaction (4). According to the results obtained from
Table 4. The order of concentration in non-elementary model the third reaction and Table 3-4 (kinetic coefficient and
order of concentration), the effect of this reaction can be
Variables of non-elementary model
considered insignificant. The temperature and catalyst
Item X6 X7 X8 X9 X10 X11 concentration are parameters which are prominent in
Value 4.994 4.993 4.996 8.029 4.574 5.195 the kinetic model. In this case, temperature and catalyst
concentration are fixed. It is very interesting that kinetic
Item X12 X13 X14 X15 X16 X17
coefficient of reactions (2) and (3) in the elementary
Value 10.00 0.0010 3.825 2.416 5.524 3.814 model was higher than the non-elementary one. These
reactions are related to NaHS components. This finding
can be attributed to the order of oxygen in reactions (2)
that the error of non-elementary model was acceptable and (3). However, the regeneration of spent caustic was
and could predict well. According to the Figure 5, the main purpose. The value of oxygen order was lower
thiosulfate had the highest concentration at the early times than the elementary model order.
of the reaction. Due to the main oxidation reactions (1),
decrease the amount of thiosulfate with time. Specially, Conclusion
by increasing the amount of oxygen and time, the sulfate The conclusions drawn from this study can be resumed
content also increases (35,36). as follows:
Oxygen participates in all five reactions, so it has more In order to regenerate and treat the spent caustic,
terms in the reaction equation. The results of elementary experiments were conducted via a catalytic air oxidation
model for oxygen (despite the relatively low error compared process in a bubble column reactor. A mathematical model
to the other compounds), show that these models has not was developed for regeneration of spent caustic in the
been able to accurately estimate the amount of oxygen in batch system and it was investigated by the experiment.
the reactor (Figure 6). The effects of oxygen concentration The temperature, catalyst concentration, pH, and air flow
on the removal of sulfur compounds is directly related are constant and reaction time from operating conditions
with evaluation of kinetics of reaction (16). is unstable.
NaHS (Figure 7) was accelerated to thiosulfate in the It was found that heterogeneous cobalt phthalocyanine is
reaction 3 and estimation was rather complex and a suitable catalyst for chemical oxidation of spent caustic
unsatisfactory, because thiosulfate is an intermediate and under highly alkaline conditions.
unstable component. The kinetics of NaHS is predicted The GA was used to predict the rate coefficient and kinetic
by non-elementary model slightly. The kinetic of the order of component concentrations. The elementary
overall reaction tends to the consumption of sulfide. The and non-elementary models are considered for the
unstable sulfur compound like hydrosulfide, is converted whole of reactions. Based on the GA method, the non-
to thiosulfate and sulfate via increasing oxygen (35). elementary model has reasonable results comparison to
Mercaptide (NaSR) is considered rather stable under the elementary one.
normal ambient temperature and pressure. The chemical Fundamentally, when the materials are produced or
oxidation of the mercaptide components to disulfides is consumed in different reactions the estimation is difficult
actually slow (37). So, an appropriate catalyst is needed to via the elementary model. The errors of the objective
increase the rate of conversion to disulfide. The disulfides function of elementary and non-elementary models was
are usually insoluble in the alkaline solution and can be 134.96% and 3.01%, respectively. It was also revealed that
separated easily. the non-elementary model is able to compute true values
According to Figure 9, the increase of reaction time with the lowest error.
helps approach the value of non-elementary model to
the experimental amount and results in a low error at the Acknowledgments
end of the reaction time. Chemical oxidation of sulfide in The authors would like to thank the University of
alkaline media is usually generated thiosulfate (38). Maragheh for their cooperation.

Environmental Health Engineering and Management Journal 2018, 5(4), 231–239 237
Karimi et al

Ethical issues 12. Abdollahi Y, Zakaria A, Sairi NA. Degradation of high level
Ethical issues (plagiarism, informed consent, misconduct, m-Cresol by zinc oxide as photocatalyst. CLEAN – Soil, Air,
data fabrication and/or falsification, double publication Water 2014; 42(9): 1292-7. doi: 10.1002/clen.201300451.
and/or submission redundancy) have been strictly 13. Golmohammadi S, Ahmadpour M, Mohammadi A,
Alinejad A, Mirzaei N, Ghaderpoori M, et al. Removal
observed by the authors.
of blue cat 41 dye from aqueous solutions with ZnO
nanoparticles in combination with US and US-H2O2
Competing interests advanced oxidation processes. Environ Health Eng Manag
The authors declare that they have no competing interest. 2016; 3(2): 107-13. doi: 10.15171/ehemj.2016.08.
14. Massa A, Hernandez S, Ansaloni S, Castellino M, Russo N,
Authors’ contributions Fino D. Enhanced electrochemical oxidation of phenol over
All authors contributed equally in all aspects of this manganese oxides under mild wet air oxidation conditions.
research like data collection, analysis, and interpretation. Electrochim Acta 2018; 273: 53-62. doi: 10.1016/j.
electacta.2018.03.178.
References 15. Kumar A, Verma N. Wet air oxidation of aqueous dichlorvos
1. Abdul-Wahab SA, Al-Hajri A, Yetilmezsoy K. Impact of the pesticide over catalytic copper-carbon nanofiberous
ambient air quality due to the dispersion of non-methane beads. Chem Eng J 2018; 351: 428-40. doi: 10.1016/j.
organic compounds from Barka landfill. Int J Environ Sci cej.2018.06.058.
Technol 2016; 13(4): 1099-108. doi: 10.1007/s13762-016- 16. Barge AS, Vaidya PD. Wet air oxidation of cresylic spent
0947-x. caustic - a model compound study over graphene oxide
2. Karimi A, Fatehifar E, Alizadeh R. Synthesis and (GO) and ruthenium/GO catalysts. J Environ Manage 2018;
characterization of nanostructured CuO/CeO2 catalysts via 212: 479-89. doi: 10.1016/j.jenvman.2018.01.066.
ultrasound assisted techniques used for selective oxidation 17. Boczkaj G, Fernandes A, Makos P. Study of different
of CO. Iranian Journal of Chemical Engineering 2013; advanced oxidation processes for wastewater treatment
10(3): 51-9. from petroleum bitumen production at basic pH. Ind
3. Ahmad W. Neutralization of spent caustic from LPG plant Eng Chem Res 2017; 56(31): 8806-14. doi: 10.1021/acs.
at preem AB Goteborg [dissertation]. Goteborg, Sweden: iecr.7b01507.
Chalmers University of Technology; 2010. 18. Ahmadpour A, Haghighi Asl A, Fallah N. Synthesis and
4. Sheu SH, Weng HS. Treatment of olefin plant spent caustic photocatalytic studies of TiO2-clinoptilolite on spent
by combination of neutralization and Fenton reaction. caustic wastewater treatment. Particul Sci Technol 2018;
Water Res 2001; 35(8): 2017-21. doi: 10.1016/S0043- 36(7): 791-8. doi: 10.1080/02726351.2017.1302534.
1354(00)00466-8. 19. Karimi A, Fatehifar E, Alizadeh R, Ahadzadeh I.
5. Debellefontaine H, Chakchouk M, Foussard JN, Tissot Regeneration of spent caustic of olefin unit in a bubble
D, Striolo P. Treatment of organic aqueous wastes: wet air column reactor: treatment and recovery optimization.
oxidation and wet peroxide oxidation. Environ Pollut 1996; Environ Prog Sustain Energy 2017; 36(2): 341-7. doi:
92(2): 155-64. doi: 10.1016/0269-7491(95)00100-X. 10.1002/ep.12433.
6. Alnaizy R. Economic analysis for wet oxidation processes 20. Sivapathasekaran C, Sen R. Performance evaluation of an
for the treatment of mixed refinery spent caustic. Environ ANN–GA aided experimental modeling and optimization
Prog 2008; 27(3): 295-301. doi: 10.1002/ep.10261. procedure for enhanced synthesis of marine biosurfactant
7. Jafarzadeh Haghighifard NA, Jorfi S, Ahmadi M, Mirali S, in a stirred tank reactor. J Chem Technol Biotechnol 2013;
Kujlu R. Treatment of mature landfill leachate by chemical 88(5): 794-9. doi: 10.1002/jctb.3900.
precipitation and Fenton advanced oxidation process. 21. Sowmiya N, Valarmathi B, Srinivasa Gupta N, Essaki
Environ Health Eng Manag 2016; 3(1): 35-40. Muthu P, Rajendran C. A novel genetic algorithm for
8. Karimi A, Fatehifar E, Alizadeh R, Ahadzadeh I. solving machine part cell formation problem considering
Regeneration and treatment of sulfidic spent caustic using alternative process plans. Mater Today Proc 2018; 5(5 Pt 2):
analytic hierarchy process. Environ Health Eng Manag 13574-84. doi: 10.1016/j.matpr.2018.02.353.
2016; 3(4): 203-8. doi: 10.15171/ehem.2016.21. 22. Gen M, Cheng R. Genetic Algorithms and Engineering
9. Diaz de Tuesta JL, Garcia-Figueruelo C, Quintanilla A, Design. Ashikaga, Japan: John Wiley & Sons, Inc; 1997. doi:
Casas JA, Rodriguez JJ. Application of high-temperature 10.1002/9780470172254.
fenton oxidation for the treatment of sulfonation plant 23. Goldberg DE. Genetic Algorithms in Search, Optimization
wastewater. J Chem Technol Biotechnol 2015; 90(10): 1839- and Machine Learning. New York: Addison-Wesley; 1989.
46. doi: 10.1002/jctb.4494. 24. Srinu Naik S, Pydi Setty Y. Optimization of parameters
10. Oh SY, Shin DS. Degradation of spent caustic by Fenton and using response surface methodology and genetic algorithm
persulfate oxidation with zero-valent iron. J Chem Technol for biological denitrification of wastewater. Int J Environ
Biotechnol 2013; 88(1): 145-52. doi: 10.1002/jctb.3876. Sci Technol 2014; 11(3): 823-30. doi: 10.1007/s13762-013-
11. Chavoshani A, Amin MM, Asgari G, Seidmohammadi A, 0266-4.
Hashemi M. Chapter 8 - Microwave/Hydrogen Peroxide 25. Kadri RL, Boctor FF. An efficient genetic algorithm to solve
Processes. In: Ameta SC, Ameta R, eds. Advanced the resource-constrained project scheduling problem with
Oxidation Processes for Waste Water Treatment. Academic transfer times: the single mode case. Eur J Oper Res 2018;
Press; 2018. p. 215-55. 265(2): 454-62. doi: 10.1016/j.ejor.2017.07.027.

238 Environmental Health Engineering and Management Journal 2018, 5(4), 231–239
 Karimi et al

26. Fatemi S, Masoori M, Bozorgmehry Boozarjomehry R. method. J Nat Gas Sci Eng 2014; 19: 62-73. doi: 10.1016/j.
Application of genetic algorithm in kinetic modeling and jngse.2014.04.015.
reaction mechanism studies. Iranian Journal of Chemistry 33. Mohammadbeigi K, Tajerian M. Demercaptanization of
and Chemical Engineering 2005; 24(4): 37-46. Distillate (DMD). Pet Coal 2004; 46(1): 17-22.
27. Balland L, Estel L, Cosmao JM, Mouhab N. A genetic 34. Jagushte MV, Mahajani VV. Insight into spent caustic
algorithm with decimal coding for the estimation of kinetic treatment: on wet oxidation of thiosulfate to sulfate.
and energetic parameters. Chemometr Intell Lab Syst 2000; J Chem Technol Biotechnol 1999; 74(5): 437-44. doi:
50(1): 121-35. doi: 10.1016/S0169-7439(99)00057-X. 10.1002/(SICI)1097-4660(199905)74:5<437::AID-
28. American Public Health Association (APHA), Eaton, JCTB63>3.0.CO;2-1.
Andrew D, Water Environment Federation (WEF), 35. Kalantari H, Nosrati M, Shojaosadati SA, Shavandi M.
American Water Works Association (AWWA). Standard Investigation of transient forms of sulfur during biological
Methods for The Examination of Water and Wastewater. treatment of spent caustic. Environ Technol 2018; 39(12):
21st ed. Washington, DC: APHA, AWWA, WEF; 2005. 1597-606. doi: 10.1080/09593330.2017.1334707.
29. Rolia E, Chakrabarti CL. Kinetics of decomposition of 36. Nosrati M, kalantari H, Shojaosadati SA, Shavandi M.
tetrathionate, trithionate, and thiosulfate in alkaline media. Optimization of factors affecting on sulfide oxidation from
Environ Sci Technol 1982; 16(12): 852-7. doi: 10.1021/ synthetic spent caustic by haloalkaliphilic Thioalkalivibrio
es00106a006. versutus by focus on sodium ion effect: application of
30. Levenspiel O. Chemical Reaction Engineering. 3rd ed. response surface methodology. Iranian Journal of Chemical
USA: John Wiley & Sons; 1999. Engineering 2017; 14(1): 74-87.
31. Soltani H, Shafiei S, Edraki J. Reactor network synthesis 37. Ward C. Method for oxidizing mercaptans and mercaptide
using coupled genetic algorithm with the quasi-linear compounds from aqueous alkaline solutions and
programming method. Chem Biochem Eng Q 2016; 30(2): hydrocarbon distillates. US 4090954 A. 1978.
199-211. doi: 10.15255/CABEQ.2014.2163. 38. Millano EF, Sorber CA. Treatment of Thiosulfate-containing
32. Behroozsarand A, Soltani H. Hydrogen plant heat exchanger wastewater in activated sludge systems. J Water Pollut
networks synthesis using coupled Genetic Algorithm-LP Control Fed 1986; 58(9): 917-23. doi: 10.2307/25043078.

Environmental Health Engineering and Management Journal 2018, 5(4), 231–239 239

View publication stats