Accepted Manuscript

Title: Comparison of the adsorption capacity of organic

compounds present in produced water with commercially
obtained walnut shell and residual biomass

Authors: Gallo-Cordova Alvaro, Silva-Gordillo Marı́a del

Mar, Muñoz Gustavo A., Arboleda-Faini Xavier, Almeida
Streitwieser Daniela

PII: S2213-3437(17)30356-1
DOI: http://dx.doi.org/doi:10.1016/j.jece.2017.07.052
Reference: JECE 1767

To appear in:

Received date: 7-4-2017

Revised date: 20-7-2017
Accepted date: 22-7-2017

Please cite this article as: Gallo-Cordova Alvaro, Silva-Gordillo Marı́a del Mar, Muñoz
Gustavo A., Arboleda-Faini Xavier, Almeida Streitwieser Daniela, Comparison of the
adsorption capacity of organic compounds present in produced water with commercially
obtained walnut shell and residual biomass, Journal of Environmental Chemical
Engineeringhttp://dx.doi.org/10.1016/j.jece.2017.07.052

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 Comparison of the adsorption capacity of organic compounds
present in produced water with commercially obtained walnut shell
and residual biomass
Gallo-Cordova, Alvaroa; Silva-Gordillo, María del Mar a; Muñoz, Gustavo A.a; Arboleda-
Faini, Xavierb; Almeida Streitwieser, Danielaa,1
a
Universidad San Francisco de Quito USFQ, Department of Chemical Engineering. Institute for
Development of Alternative Materials and Energies IDEMA. Av. Diego de Robles y Vía
Interoceánica, Quito, Ecuador, EC170901
b
AF Hidrotecnología Cia. Ltda., Calle D y Calle del Establo, Urban Tower, ofic. 304, Quito,
Ecuador, EC170157
1
Corresponding author: dalmeida@usfq.edu.ec / Tel. Number: +593 22971700 ext.:1420#
Graphical Abstract

Highlights:

1
 Utilization of residual biomass as adsorbent for purification of produced water has been proven as
feasible.
 Langmuir adsorption isotherm fits best to the adsorption equilibrium data of organic compounds in
residual biomass
 The pseudo – first order kinetic model adjusts well to the adsorption of organic compounds in
walnut shell and in palm shell used as adsorbents.

Abstract
The effectiveness of the adsorption of organic compounds in produced water with different residual
biomass sources has been investigated. Adsorption capacities have been analyzed with palm shell,
orange peel, banana peel, passion fruit peel, cocoa bean and sawdust; and compared to the commercial
adsorbent walnut shell. All adsorbents undergo a pretreatment, prior to the adsorption, consisting of
washing, drying and thermal conditioning. For the investigation of the adsorption capacity of the
different adsorbents, synthetic produced water has been prepared with different organic compounds
and expressed as the concentration of Chemical Oxygen Demand (COD). The adsorption capacities of
the different adsorbents are expressed as the removal percentage of oxidizable contaminants,
presented as COD concentration in the synthetic produced water. The adsorption analyses with the
synthetic produced water in a batch experimental procedure show that only palm shell and sawdust
actually adsorb the organic compounds in the synthetic water. Three isotherm models were considered:
Langmuir, Freundlich and Temkin. Among these three models, the Langmuir model describes best the
experimental data for walnut shell, palm shell and sawdust. The maximum adsorption capacity was
determined and sawdust presented a maximum value of contaminant removal of 33 mg/g of dry
adsorbent, while walnut shell and palm shell presented a contaminant removal of 4.9 and 5.6 mg/g of
dry adsorbent, respectively. Kinetics results were best described by pseudo-first order potential law
model for the experiments carried out with the walnut shell and palm shell. Breakthrough experiments
show that palm shell will saturate earlier than walnut shell. CB Cocoa beans

COD Chemical Oxygen Demand, mg/L

OP Orange peel
oTS Organic Total Solids, %
PP Passion fruit peel
PS Palm shell
SD Sawdust
SPW Synthetic Produced Water
TS Total Solids, %
WS Walnut shell

Abbreviations

BP Banana peel

Keywords: produced water; walnut shell; adsorbent; residual biomass; adsorption kinetics;
isotherm; adsorbent breakthrough

2
1 Introduction
Oil and gas production tend to generate large quantities of contaminated water known as produced
water. This water is composed of formation water, the water naturally present in the reservoir, and
injection water, that could have been previously injected in the reservoir [1]. Both formation and
injection water eventually rise to the surface and are extracted at the wellhead along with gas and oil.
The volume of water needed for injection, depends on the type of deposit and its production
mechanism. In most countries legal regulations for injecting wells are increasingly stringent to protect
fresh water aquifers [2]. Due to the large amount of contaminants in produced water, it cannot be
discharged directly into natural water cycles or reinjected back into the well [3]. As oil extraction rises
the amount of reinjected water increases considerably, in order to maintain the underground reservoir
pressure. Because of the great importance of not wasting water resources, water treatment for
reinjection has become crucial for oil-producing companies [4]. For this reason, it is necessary to
develop new water treatment technologies, which will reduce the difficulties during water injection in
the bottom of the wells or for the disposal of produced water [5]. Finding the most appropriate
technology for water treatment production depends on several factors, such as disposal methods,
environmental impact and economic investment [6]. The produced water treatment is based on a
sequence of steps according to the oil and solids amount present in the water. The first stage is
characterized by the separation of hydrocarbons by flotation units. In order to achieve maximum purity
in the first stage, a highly effective filtration process is employed [7]. Filters based on organic adsorbents
are particularly suitable for this application, ensuring the elimination of up to 98 % of contaminants in
the feed water [7].

Worldwide, 140 million metric tons of biomass are generated each year by agricultural activities. This
volume of biomass can be converted into an enormous amount of energy and raw material [8], but this
biomass can also be used as adsorbents for filtration processes using adsorption. The main advantages
of filtration using organic adsorbents include the use of large amounts of residual biomass, which leads
to obtaining low cost raw materials, high efficiency for contaminant removal and minimizing the use of
chemical or biological products [9,10]. In this project, six different residual biomasses were used to
determine its potential in removing organic compounds present in produced water as an alternative to
the imported adsorbent walnut shell (WS). Currently, WS is used as an adsorbent in produced water
treatment plants internationally. The experiments were carried out by analyzing the removal percentage
of oxidizable contaminants expressed as Chemical Oxygen Demand (COD) and calculating equilibrium
adsorption isotherms and adsorption kinetics.

2 Materials and methods

2.1 Adsorbent selection, pretreatment and characterization
Walnut shell (WS), palm shell (PS), orange peel (OP), banana peel (BP), passion fruit peel (PP), cocoa
beans (CB), and sawdust (SD) were used as adsorbents for organic compounds in produced water. WS is
the commercial product obtained from the local importer AF Hidrotecnología, which consists only of the
shredded shells with the total particle size ranging between 0.3 and 1.18 mm, where 60% of the
particles lie between 0.3 and 0.6 mm. The same particle size was used for the other adsorbents. The
alternative adsorbents were selected considering the high local availability and the possibility to collect
them as residues from agribusinesses separately. PS was collected from a local center of palm oil

3
extraction, while SD was obtained from a local timber. CB was obtained as discarded material from a
local cocoa producer, due to non-conformities in size, form or damages. And OP, BP and PP were
obtained as residues from a local fruit processing industry. The different types of residual biomass
obtained locally were dried at 100 °C for 24 h until constant weight was obtained. Subsequently the
dried biomass were shredded and sieved to a particle size between 0.3 and 0.6 mm. The pretreatment
was finished by soaking and washing the adsorbents with distilled water until no dust remains in the
adsorbent. In the case of the PS, OP, BP, PP and CB an additional pretreatment step was performed by
heating the adsorbents at 180 °C for up to 3 hours, until constant weight was obtained.

The residual biomasses, used as adsorbents in this study, were characterized and compared to the
commercial adsorbent WS. The characterization was first carried out by determining the WS solid
gradation so the particle size for the other residual biomasses can be selected. The amount of total
solids present in PS, OP, BP, PP, CB and SD were measured by the ASTM E1756 method [11], where a dry
sample is obtained in a Thermo/Precision Scientific 25EM oven. The amount of ashes and total volatile
organic compounds were determined by the calcination of the dry samples in a Thermolyne 48000
furnace heated to 550 °C for 8 h according to the APHA Standard Methods for the Examination of Water
and Wastewater [12]. Bulk density and real density were also measured for all the residual biomasses by
the ASTM D7481 – 09 [13], so that void volume and uniformity coefficient could be calculated. Finally,
the adsorbents’ surfaces were observed using a MIRA3 TESCAN Scanning Electron Microscope.

2.2 Characterization methods and preparation of produced and synthetic water

The actual produced water was obtained from an oil extraction camp at the Amazon region of Ecuador.
The organic contamination in produced water was measured by Chemical Oxygen Demand (COD), which
shows the amount of oxygen that can be consumed by reactions in a specific volume and is a
representation of the degradable material present in a sample [14]. To obtain the COD value the water
sample was blended with H2SO4 98% (w/w) and a potassium dichromate solution and digested according
to the Manganese III Digestion Method, U.S. Patent 5,556,787 [15]. The samples were digested using a
DRB 200 digester heated to 150 °C for 2 h, so they can be measured in a HACH DR 890 colorimeter.

The composition of produced water was also analyzed by a Shimadzu Gas Chromatography - Mass
Spectroscopy (GC-MS) model QP2010 Ultra Plus (Shimadzu, Japan) equipped with an
autosampler/autoinjector AOC 20i-s and operated with the GCMS Solution software (Version 4.11 SU2).
A Restek silica column model SH-Rxi-5Sil MS has been used. The method used for the GC-MS analyses
are described by Vargas et al [16]. For the injection 1 l water sample needed to be pre- filtered with a
0.45 m PTFE filter.

The average COD of the produced water sample was 1319.36 mg/L ± 4.18. The results obtained by the
GC-MS analysis showed that produced water has compounds with large and lineal carbon chains with
carboxylic functional groups, as presented in section 4.2. Therefore, for the preparation of the synthetic
produced water (SPW), 3 % gasoline, 10 % oleic acid, 2 % lubricant motor oil and 85 % of distillated
water were used so it presents a similar COD value to the collected produced water. The contaminants
were mixed for 1 h and then, using a sedimentation funnel, the mixture was left to rest for another
hour.

2.3 Experimental setup for batch adsorption experiments

4
The batch adsorption experiments were carried out at room temperature (20 ºC ± 1ºC) in a 150 mL
Erlenmeyer containing the produced water and the adsorbent under agitation. To increase the contact
time, magnetic agitation and a shaker were used. In both cases, mechanical degradation of the
adsorbents was observed after a certain time.

The adsorption processes generally are affected by the adsorbent dosage. To obtain the maximum
adsorption capacity it is necessary to determine the optimum dosage of adsorbent. By obtaining the
effect of the adsorbent dosage, the amount of adsorbent needed to remove certain amount of
contaminants can be calculated [17]. The experimental setup was carried out by mixing 4 g of adsorbent
with 40.0 mL of synthetic produced water (SPW) with a known COD concentration. Because of SD
density, the batch experiments using this adsorbent were carried out using 0.2 g of SD in 40.0 mL of
SPW. This value was chosen because in that case the volumetric ratio is kept constant at 4.5 mL of
adsorbent in 40 mL of SPW.

In this study the commercial adsorbent used for comparison was WS. The maximum value of WS dosage
was also used for PS, OP, BP, PP, and CB. The effect of the amount of adsorbent was studied, ranging
dosages from 0 to 10 g/40.0 mL with 1300 mgCOD /L of SPW. Also, the effect of the contact time was
determined by agitating the mixtures for different times between 1 to 120 min, and then filtering the
mixture so COD can be measured. The Langmuir, Freundlich and Temkin isotherm models were used to
describe the adsorption process. Pseudo first order and pseudo second order models were used to
determine the adsorption kinetic mechanisms. The experiments were performed by triplicate so the
results have statistical relevance.

2.4 Filtration system design for breakthrough experimentation

In order to set a good comparison between the commercial adsorbent and the best alternative
adsorbent, it is necessary to determine the saturation time for the residual biomasses used [18]. The
breakthrough curves for the best adsorbents were obtained. These experiments were settled in a
laboratory packed bed filtration system with downward flow as shown in Figure 1.

The breakthrough curves were carried out as a continuous operation with a residence time for the SPW
in the filtration system of 15 minutes. The packed bed occupied 50 % of the filter. Experiments were
performed working with different pressures values at the inlet and outlet, so channeling could be
avoided. Samples of treated water were collected every minute for an hour. COD of the samples was
analyzed and the breakthrough time for the adsorbents was determined.

3 Theory
3.1 Adsorption capacity
The adsorption capacity of the different adsorbents was obtained by different batch experiments with
synthetic produced water. The amount of organic contaminants adsorbed per gram of adsorbent was
calculated from equation (1) [19].

qe = (C0 - Ce)V/m (1)

Where qe is the adsorption capacity at equilibrium, in mgCOD /g of adsorbent; C0 is the initial

concentration of contaminants in SPW, in mgCOD /L; Ce is the equilibrium concentration of contaminants
in SPW, in mgCOD/L; m is the dry weight of adsorbent, in g; and V is the volume of SPW, in L. The
5
percentage of removal of the contaminants present in SPW was calculated using the following equation
(2) [20].

% Removal = (C0 – Ce) 100/C0 (2)

3.2 Adsorption isotherms

To understand the adsorption process, isotherm models have been described. An isotherm is a term
used to describe the equilibrium curves in an adsorption process [21]. When talking about predictive
modeling for design and analysis of adsorption systems, the equilibrium isotherms have a significant
role. The experimental data obtained in a specific study may fit under one set of conditions, ergo an
isotherm model, but fail under a different one [22]. In this study three isotherm models were
considered: Langmuir, Freundlich and Temkin. Table 1 shows the non-linear and linear equation for the
models mentioned. In all models the dependence of the equilibrium adsorption capacity, qe in mg COD /g
of adsorbent, is given as a function of the equilibrium concentration of contaminants, Ce in mg/L. It is
important to consider each of the models presented, because they consider different parameters to
explain the behavior of the adsorption process.

The Langmuir adsorption model is the most common model used to quantify the amount of
contaminants present in the adsorbent as a function of the partial pressure or the concentration at a
given temperature [22,23]. This model assumes that the adsorbent is in contact with a solution that
contains adsorbate attracted to the surface and that the surface possesses a specific number of active
sites where the molecules of solute can be adsorbed [24,25] Thus, this model assumes that the
adsorption occurs over a homogenous adsorbent surface and a monolayer coverage of adsorbate takes
place [22]. The Langmuir isotherm equation presented in Table 1 relates the equilibrium adsorption
capacity, qe, with the equilibrium concentration, Ce. With this equation the maximum adsorption
capacity, qm in mgCOD /g adsorbent, and Langmuir constant, b0 in L/mg, can be obtained.

The Freundlich isotherm model is given by an empirical equation, which can be used when the
adsorption involves systems with heterogeneous surfaces, where there is no adsorption limited levels
[26,27]. This model relates the contaminants concentration present in the adsorbent surface, with the
solute concentration present in the liquid in contact [28]. As presented in the equations of Table 1 the
parameters related to the qe for this case are: the adsorbent adsorption capacity k in (mgCOD/g
adsorbent) (L/mgCOD)1/n; and the heterogeneity factor, n. The heterogeneity factor obtained is a measure
of the adsorption intensity or the surface heterogeneity. When the value of the slope gets near zero, it
means that the process is more heterogeneous [29].

Due to the possible adsorption heat interaction, the Temkin isotherm model is considered. This model
assumes that the energy of the molecular adsorption is characterized by a uniform distribution of the
bond energies [30]. This means that the heat of adsorption of the molecules present in the surface layer
would decrease with the coverage in a linear way [25]. As presented in Table 1, the discussed isotherm
considers: the ideal gases constant, R as 8.314 J/mole K; the temperature at which adsorption takes
place, T in K; Temkin constant related to the adsorption heat, bt in J/mol, and Temkin isotherm constant,
kt in L/g [31].

Table 1: Isotherm models equations

6
 Isotherm Equation Linearized equation Plot

𝑞𝑚 𝑏0 𝐶𝑒 𝐶𝑒 1 1 𝐶𝑒
𝑞𝑒 = = + 𝐶 𝑣𝑠 𝐶𝑒
Langmuir 1 + 𝑏0 𝐶𝑒 𝑞𝑒 𝑏0 𝑞𝑚 𝑞𝑚 𝑒 𝑞𝑒

1
𝑞𝑒 = 𝐾𝑓 𝐶𝑒 1/𝑛 𝑙𝑛𝑞𝑒 = 𝑙𝑛𝐾𝑓 + 𝑙𝑛𝐶𝑒 𝑙𝑛𝑞𝑒 𝑣𝑠 𝑙𝑛𝐶𝑒
Freundlich 𝑛

𝑞𝑒
𝑅𝑇
= 𝑙𝑛𝐾𝑇
𝑏𝑇
𝑅𝑇
+ 𝑙𝑛𝐶𝑒
𝑏𝑇

𝑞𝑒
Temkin 𝑅𝑇
= ln⁡(𝐾𝑇 𝐶𝑒 )
𝑏𝑇

𝑞𝑒 𝑣𝑠 𝑙𝑛𝐶𝑒

Table 2: Kinetic models equations

Kinetic model Equation Linearized equation Plot

< 𝐼𝑛𝑙𝑖𝑛𝑒𝑆ℎ𝑎𝑝𝑒6 >
< 𝐼𝑛𝑙𝑖𝑛𝑒𝑆ℎ𝑎𝑝𝑒5 >
< 𝐼𝑛𝑙𝑖𝑛𝑒𝑆ℎ𝑎𝑝𝑒4 >

Pseudo-first order

1 1 𝑡
= 2 + 1
𝑑𝑞𝑡 2 𝑞𝑡 𝑞𝑒 𝑘2 𝑞𝑒
= 𝑘2 (𝑞𝑒 − 𝑞𝑡 ) 𝑣𝑠 𝑡
Pseudo-second order 𝑑𝑡 𝑞𝑡

3.3 Adsorption kinetics

7
In order to understand the mechanism of adsorption and the possible rate – determining step, including
mass transport and chemical reaction, kinetics models are analyzed. In this study the models that are
used to understand the adsorption of organic compounds with residual biomass are the pseudo-first
order and pseudo-second order equations. These models are illustrated in Table 2 with the
corresponding non linearized expressions; where k1 and k2 are the kinetic constants for each model. The
rate of change of the adsorption capacity, dqt /dt, is expressed as a function of the difference of
adsorption capacity, qt, and the contact time, t, for the reaction orders 1 and 2. The non-linear
equations for both models can be linearized by integration, where the boundary conditions are qt = 0 in
t = 0 and qt = qt in t= t. In the case of the pseudo-first order equation, the constant k1 and the value of
the adsorption capacity in equilibrium, qe, are obtained by plotting ln (qe – qt) vs t. On the other hand,
the pseudo-second order model is based on the adsorption of solid phases [21]. A plot t/qt versus t gives
a line from which the constants k2 y qe can be obtained.

4 Results and discussion

4.1 Characterization of the adsorbents
Each adsorbent was characterized prior to determining its adsorption properties. The results of the
characterization are shown in Table 3. Humidity is a very relevant parameter, since the solid with lower
water content will need more time to get saturated with water and so start adsorbing the organic
compounds. Also, the organic total solids (oTS) show the organic compounds that are present in the
residual biomasses. Adsorbents with higher amounts of oTS may show a lower adsorption capacity,
since the adsorbent will get saturated faster. For this reason, all materials with high oTS percentages will
need an additional pretreatment in order to reduce the amount of oils and organic compounds already
present in the adsorbents. Figure 2 compares basic composition of the different biomass adsorbents
divided into organic ashes and water fraction. In Table 3, also bulk density and real density are
presented, as well as uniformity coefficient and particle size.

Table 3: Characterization of the different adsorbents

Characterization of adsorbents
Unit WS PS OP BP PP CB SD
Humidity % 5.113 3.987 6.231 5.332 8.453 4.245 1.354
Ashes % 52.119 45.779 13.427 17.323 2.984 20.319 81.079
oTS % 42.768 50.234 80.342 77.345 88.563 75.436 17.567
Real
g/mL 0.698 0.754 0.709 0.632 0.453 0.509 0.018
density
Bulk
g/mL 0.875 0.899 0.854 0.785 0.653 0.621 0.045
density
Uniformit
y
- 1.75 1.82 1.732 3.713 1.345 2.843 1.872
coefficien
t
Particle
mm 0.3-1.18 0.3-0.6 0.3-0.6 0.3-0.6 0.3-0.6 0.3-0.6 0.3-0.6
size
WS: Walnut shell, PS: Palm shell, OP: Orange peel, BP: Banana peel, PP: Passion fruit peel, CB: Cocoa beans, SD: sawdust

8
All adsorbents used in this work, excepting PP and CB, were characterized by scanning electron
microscopy. Figure 3 shows that all adsorbents have an irregular surface, where WS and PS have a very
similar structure matrix and small pores can be identified at different sections, which indicate high
surface areas for adsorption, while in SD the broken fibers can be identified as smooth and thin layers.
The images of OP and BP have been chosen as representatives for the biomasses that contained high
amounts of organic material, showing a high amorphous and irregular structure without visible pores.

4.2 Characterization of produced water at different stages of the process

The produced water was characterized by GC-MS. The main components present in the produced water
were determined by this method. According to Figure 4 the produced water contains organic
compounds with long carbon chains such as octadecanoic acid and hexadecanoic acid. Based on these
results, synthetic produced water was prepared using similar compounds with long carbon chains. As
mentioned before, oleic acid, gasoline, and lubricant motor oil were used to prepare the produced
water. Both water samples, original produced water and synthetic produced water (SPW), were
analyzed by GC-MS and compared. As shown in Figure 5, octadecanoic acid is presented in the synthetic
produced water as is oleic acid with a C18 carbon chain.

Finally, the synthetic produced water was analyzed by GC-MS after the adsorption process with the
different adsorbents, in order to determine its capability to remove the organic compounds. In Figure 6
the chromatogram after adsorption of synthetic produced water with the commercial WS is presented.
The same peaks can be observed, but with a significantly lower intensity, showing a reduction in its
concentration. For the alternative adsorbents PS and SD, similar chromatograms have been obtained
showing a similar decrease in the peaks for both PS and SD, indicating that the residual biomasses used
as adsorbents have the same affinity for the long chain organic compounds used in SPW.

4.3 Definition of the Experimental Setup

The experimental setup was defined using the commercial walnut shell. The adequate parameter range
for adsorbent dosage, contact time and initial concentration are defined by varying each one
systematically. Once the optimum parameters are obtained for walnut shell, the alternative adsorbent
can be investigated.

The effect of WS dosage in the reduction in the content of organic contaminants in SPW is shown in
Figure 7. The reduction in the content of the organic contaminants in the SPW is represented as the
reduction in the value of the Chemical Oxygen Demand. The adsorption capacity of the adsorbent is
expressed as the removal percentage of oxidizable contaminants, presented as COD concentration in
the synthetic produced water. As expected, the percentage removal tends to increase while the dosage
9
of adsorbent is higher. The maximum adsorption capacity is reached at a dosage of WS between 5.0 –
8.0 g with 60 % of removal at an initial COD concentration of 400 mg/L. Therefore, the optimal dosage
defined for the batch experiments is set to 4.0 g of adsorbent in 40 mL of SPW. The relation adsorbent –
water is defined as 1:10. The only exception is the case of the SD, were the amount of adsorbent used
was of 0.2 g adsorbent /40 mL of SPW since the density of SD is too low. The results with the different
adsorbents are still comparable, since the volumetric ratio in all cases remained constant at a ratio of
4.45 – 4.55 ml of adsorbent / 40 mL of SPW.

Next the effect of contact time was analyzed. This parameter provides information on how fast the
adsorption process is taking place. Figure 8 shows the effect of contact time for the removal of
contaminants in SPW by WS. The maximum removal was reached within 10 minutes, regardless of the
agitation type used. It can be observed that at higher contact times the particles started to be shredded
by the agitation, affecting the concentration of COD in the synthetic produced water. Therefore, for
further investigations on the effect of the initial concentration on the percentage removal, a maximum
agitation time of 10 minutes was used. The results are confirmed in Figure 9, where the percentage
removal is presented as function of time for different initial COD concentrations.

Finally, the salt content in the synthetic produced water is also considered, since it can interfere with
the adsorption capacity of the adsorbent. According to Fakhru’l-Razi et al [2], a common salinity
concentration in produced water is 55 g/L NaCl equivalent. All salinity studies were accomplished using
WS at constant COD concentration and varying the initial salt concentration within the reported values,
as presented in Table 4. As shown, the analysis demonstrates that salinity does not affect the
percentage removal of oxidizable contaminants expressed as COD.

Table 4: Salinity studies with an initial COD concentration of 1432 mg/L

Percentage Removal as function of salt concentration
Parameter Unit Value
NaCl g/L 0 30 40 50 60 70
COD mg/L 998 1001 1064 999 1026 1041
Removal % 30.31 30.09 25.69 30.24 28.35 27.30

After the adsorption experiments with different salt concentrations, the adsorbent did not present any
visual change in its surface, According to the COD values presented in Table 4, it can be concluded that
for this study salt content in produced water does not affect adsorption. Yet, more analyses should be
made at higher salt concentration in order to verify its relevance in the adsorption process.

The analyses carried out with walnut shell provided the necessary information to perform the following
experiments with the alternative adsorbents at optimal conditions. For the batch experiments the
optimum adsorbent dosage is set as 4 g in 40 mL of SPW, agitated during 10 minutes. Also the results
showed that neither salt concentration affects the adsorption, nor the adsorbent characteristics.

4.4 Comparison of the adsorption capacity of different residual biomasses

For the adsorption analyses of the different adsorbents, the optimal adsorbent dosage was used. COD
was measured from samples taken after 1, 3, 5 and 10 minutes of contact time. As shown in Table 5, OP,
10
BP, PP, and CB did not adsorb even after the thermal pretreatment at 180 °C. Due to the large amounts
of soluble compounds, such as tannins, coloring agents, sugars and resins present in mentioned
biomasses, the COD for this biomasses increases considerably after adsorption, because the mentioned
components are solubilized in the produced water altering its color and COD concentration [28,32].
Thus, produced water presented a higher COD value ending up more contaminated rather than purified.
Some studies have shown that these substrates are capable of adsorbing organic compounds, as long as an
alternative activation method is used, such as heat or acid wash [33–36].

Table 5: Adsorption analysis for all adsorbents at an optimal contact time (5 min)
COD, mg/L WS PS SD OP BP PP CB
initial 377.00 488.00 445.00 464.00 464.00 464.00 464.00
final 243.33 350.35 342.51 1122.23 1072.54 993.53 1024.68

As indicated in Table 5, the COD for WS, PS and SD, decreases after adsorption. For this reason,
following analyses were carried out using only WS, PS and SD. The effect of initial concentration using
PS and SD was studied and compared to the results obtained for WS as shown in Figure 10. The removal
percentage decreases as the initial concentration increases. This figure reveals that the amount of
organic compounds adsorbed per unit of adsorbent mass increases as the initial concentration increases
from 300 to 1400 mg/L. When the initial concentration of contaminant increases, the maximum
adsorption capacity increases as well, due to a rise in the mass transfer caused by the higher
contaminants concentration [37].

The highest removal obtained was of 55 % using WS as adsorbent, while the lowest removal was 25 %
obtained in the experiments with the SD. Nevertheless, these results do not give any information
regarding the maximum adsorption capacity of the different adsorbents. Figure 11 shows the
percentage of removal obtained as function of time, varying the initial concentration of oxidizable
contaminants. As it can be observed, PS presents higher percentage removal than WS. However, WS
presented the highest removal as shown before.

The variation in the adsorption equilibrium capacity according to the initial concentration of SPW and
contact time is shown in Figure 12 for WS, PS and SD, respectively. In the case of PS the equilibrium is
reached when the initial concentration is 1404 mg/L, this value is higher than the concentration used
when working with WS. In both cases a fast adsorption of contaminants occurred. When working with
lower initial concentration of contaminants, during the first 5 minutes the percentage removal was
approximately 50 % for WS and 35 % for PS. The effect of contact time using SD showed that the
maximum removal time was reached at a contact time of 10 minutes. After this time equilibrium is
reached and no further adsorption occurs, as shown in Figure 12 c). The maximum removal percentage
with SD was 25 %, obtained at an initial concentration of 445 mg/L.

4.5 Modelling of the Adsorption Isotherms

Three different isotherms have been modelled using the results of the equilibrium adsorption capacity
(qe) of each adsorbent. A lineal plot for each isotherm equation is performed in order to obtain the
11
constant for each model and to verify which data adjust better to the plot. After comparing the R2 of
each plot, we can determine that Langmuir isotherm best describes the experimental data for walnut
shell, palm shell and sawdust. Therefore, it can be assumed that the adsorption takes place in a
monolayer. In specific homogenous sites, no more than one molecule can occupy a specific site. Hence,
no further adsorption can be accomplished [22]. As can be seen in Table 6, the maximum adsorption
capacity (qm) for walnut shell is 5.0384 mg/ g, for palm shell is 4.9240 mg/ g and for sawdust it is
32.7869 mg/ g. With these results, sawdust presented the best characteristics for adsorption. However,
the very low density of SD can affect the filtration since it could cause channeling and flotation. Finally,
the nonlinear isotherms for WS, PS and SD were drawn and shown in Figure 13 a), b) and c),
respectively. The three figures show how the experimental data fits the Langmuir isotherm better than
the other two models. The coefficients and constants obtained for each isotherm from the slope and the
intersection of the linear plots of each isotherms are presented in Table 6.

Table 6: Calculated coefficients for the isotherms with each adsorbent

Langmuir Freundlich Temkin

WS
b0 0.0013 L/mg Kf 0.0561 (mg/g)(L/g)1/n bt 20.779 J/mol
qm 5.6380 mg/gS 1/n 0.5700 - Kt 2.372 L/g
2 2 2
R 0.9861 - R 0.9972 - R 0.9851 -
PS
b0 0.0042 L/mg Kf 0.1870 (mg/g)(L/g)1/n bt 21.5760 J/mol
qm 4.9240 mg/gS 1/n 0.4690 - Kt 2.4520 L/g
R2 0.9994 - R2 0.8680 - R2 0.9956 -
SD
b0 0.0033 L/mg Kf 2.2129 (mg/g)(L/g)1/n bt 7.3192 J/mol
qm 32.7869 mg/gS 1/n 2.8329 - Kt 0.0313 L/g
R2 0.9896 - R2 0.9265 - R2 0.9416 -

4.6 Kinetics studies

The equations presented in Table 2 for the first order and second order kinetics models are fitted to the
adsorption capacity with the experimental data by linear regression analysis. Batch experiments with WS
and PS were carried out to obtain the kinetics information and the effect of contact time. The experiments
with SD were not used for the kinetic studies, because of the low density and difficulties in adjustment of
the data with this adsorbent. The pseudo-second order rate expression was first used with the experimental
data from the WS and PS experiments. For this model no fitting to the curve could be observed with any of
the adsorbents used. The pseudo-first order model was also applied to the experimental data. As shown in
the adsorption rates for both adsorbents fit very well to the pseudo-first order model with a R2 greater than
0.97 for all the variations in the initial concentration of SPW. The fitting of the data is shown in Figure 14.
The kinetics constants are shown in Table 7.

12
Table 7: Experimentally obtained kinetics parameters for pseudo-first and pseudo-second order equations
using PS and WS

Adsorbent PS WS
C0 , mg / L 1404 788 488 368 1296 780 490 377
Pseudo-first order
qe , mg / g 2.8417 2.0269 1.5110 1.2843 3.9385 3.6103 2.4298 2.1111
k1 0.3652 0.3952 0.4808 0.2647 0.1991 0.2054 0.1573 0.1442
2
R 0.9919 0.9959 0.9930 0.9773 0.9992 0.9995 0.9993 0.9992
Pseudo-second order
qe , mg / g -21.7391 -11.3636 -5.1493 -15.3609 -2.9515 -4.5475 -10.0908 -10.0908
k2 0.0029 0.0270 0.0017 0.0076 0.0343 0.0211 0.0089 0.0089
R2 0.6985 0.6208 0.7458 0.6600 0.6616 0.6616 0.6671 0.6392

4.7 Breakthrough Curves

Finally, to define if PS and SW are able to replace WS as an adsorbent for the removal of contaminants in
produced water, it is important to determine how long the adsorbent is going to work at its full capacity.
The breakthrough curves were obtained for WS and PS. SD was not considered for this experimentation,
due to its low density it was difficult to fill the solid filter.

The breakthrough curves were obtained for WS and PS for one single wash and they are presented in
Figure 15. As expected, the saturation time for PS is lower than for WS, because the maximum
adsorption capacity of PS is higher. Therefore, the saturation point is going to be reached faster. For WS
the breakthrough of the material started after 15 minutes, while for PS it started after approximately 8
minutes.

5 Conclusion
This study shows that residual biomass can be used as adsorbents for the removal of organic
compounds present in produced water. For this investigation palm shell, saw dust, orange peel, banana
peel, passion fruit peel, and cocoa beans were used as possible adsorbents and compared to the
commercial product walnut shell. The synthetic produced water for the adsorption of organic
compounds was prepared based on an original sample obtained from an oilfield located in Ecuador’s
Amazon region. For the adsorption experiments the effect of contact time, adsorbent dosage, initial
COD concentration and salinity were analyzed first with the walnut shell adsorbent. Then the
percentage removal capacity of the other adsorbents were analyzed. The results obtained with the
13
adsorption analyses show that even though residual biomasses are good adsorbents, not all of them can
be used without a pretreatment. It was observed that orange peel, banana peel, passion fruit peel and
cocoa beans were not able to perform a good adsorption in this study. As it was mentioned, some
adsorbents are capable of adsorbing organic compounds, as long as an activation method is used, such
as heat or acid was. Therefore, future investigations should be performed on the pre-treatment
methods.

In this study the batch experiments have been performed with WS, PS and DS. The three adsorbents
were used for modeling the adsorption isotherms. The results suggest that all three models: Langmuir,
Freundlich and Temkin, presented a good fitting for the adsorption of the contaminants. Though,
Langmuir was the model which best describes the experimental data in all three cases. A R2 greater than
0.9 was obtained for the three linear plots. The maximum adsorption capacity was determined using the
constants values obtained from the linear plots and the Langmuir equation. Sawdust had the highest
value. However, this material has very low density which can cause several problems during the
filtration process. In order for this adsorbent to be used in industrial processes, the filtration system
must include a packed bed where SD can be compressed so channeling and flotation could be avoided.

Kinetics models were performed to determine the adsorption rate for the two adsorbents: WS and PS.
The results showed that both, walnut shell and palm shell, are better described by a pseudo-first order
reaction. With these parameters, a more substantial study could be carried out, including the
regeneration or backwashing process of the biomass to analyze its stability and durability. Breakthrough
curves showed that palm shell saturate faster than walnut shell. Breakthrough experiments were not
performed using sawdust because its low density and also the filter did not present a fixation system for
the adsorbent to be compressed in the packed bed.

6 Acknowledgments
The authors thanks AF Hidrotecnología for the project description and the provision of the raw materials
during the development of the project and Universidad San Francisco de Quito for the continuous
support to the investigations performed at IDEMA.
This work was supported by AF Hidrotecnología through the Cooperation Agreement in Research in
Filtrations Systems 2015.

7 Reference
[1] R.S. Kraus, Petroleo: Prospeccion y Perforacion, in: Encicl. Salud Y Segur. En El Trab., OIT, 2012:
pp. 1–16.
[2] A. Fakhru’l-Razi, A. Pendashteh, L.C. Abdullah, D.R.A. Biak, S.S. Madaeni, Z.Z. Abidin, Review of
technologies for oil and gas produced water treatment, J. Hazard. Mater. 170 (2009) 530–551.
[3] R. Farajzadeh, Produced Water Re-Injection ( PWRI ) An Experimental Investigation into Internal
Filtration and External Cake Build up, (2004).
[4] J.N. Cabrera Mármol, F.E. Cabrera Ochoa, J.P. Delgado Chancay, I.R. Gallegos Orta, “Análisis y
Propuestas de Mejora de Proyectos de Inyección de Agua Ejecutados en Reservorios de Diversas
Características,” Artículos Tesis Grado - FICT. 9 Mayo (2012) 2–9.
14
[5] E.T. Igunnu, G.Z. Chen, Produced water treatment technologies, Int. J. Low-Carbon Technol. 9
(2014) 157–177.
[6] R. Al-maamari, M. Sueyoshi, Flotation Filtration and Adsorption Pilot Trials for Oilfield Produced
Water Treatment, Soc. Pet. Eng. (2012) 56–64.
[7] J. Arthur, B. Langhus, C. Patel, Technical Summary of Oil & Gas Produced Water Treatment
Technologies, ALL Consult. LLC. (2005) 1–53.
[8] J. United Nations Environmental Programme Division of Technology, Industry and Economics
International Environmental Technology Centre Osaka/Shiga, C onverting W aste A gricultural B
iomass into a R esource, United Nations Environ. Program. (2009) 1–437.
[9] M.S.Z. Lopicic, J. Milojkovic, C. Lacnjevac, M. Mihajlovici, M. Kostic, A. Petrovici, Biomass waste
material as potential adsorbent for sequestering pollutants, Sci. Pap. 53 (2012) 231–237.
[10] I. Ahmad, N. Ali, Y. Jamal, Treatment of Domestic Wastewater by Natural Adsorbents Using
Multimedia Filter Technology, 4 (2016) 164–167.
[11] ASTM E1756-01, Standard Test Method for Determination of Total Solids in Biomass, 2001.
[12] L.S. Clesceri, Standard Methods for the Examination of Water and Wastewater, 1998.
[13] ASTM D7481-09, Standard Test Methods for Determining Loose and Tapped Bulk Densities of
Powders using a Graduated Cylinder, 2009. doi:10.1520/D7481-09.1.
[14] L.S.C. Eugene W. Rice, Rodger B. Baird, Andrew D. Eaton, Standard Methods for the Examination
of Water and Wastewater, Am. Water Work. Assoc. Public Work. Assoc. Environ. Fed. (2012)
1469.
[15] Hach, Datalogging Colorimeter Handbook DR890, 2007.
[16] D.C. Vargas, M.B. Alvarez, A. Hidrobo Portilla, K.M. Van Geem, D. Almeida Streitwieser, Kinetic
Study of the Thermal and Catalytic Cracking of Waste Motor Oil to Diesel-like Fuels, Energy and
Fuels. 30 (2016) 9712–9720.
[17] T.W. Seow, C.K. Lim, Removal of Dye by Adsorption : A Review, 11 (2016) 2675–2679.
[18] H. Wang, N. Persaud, T. Lin, S. Mostaghimi, Y. Pachepsky, L. Zelazny, Describing and Predicting
Breakthrough Curves for non-Reactive Solute Transport in Statistically Homogeneous Porous
Media, Virginia Polytech. Inst. State Univ. (2002).
[19] B. Singha, S.K. Das, Adsorptive removal of Cu(II) from aqueous solution and industrial effluent
using natural/agricultural wastes, Colloids Surfaces B Biointerfaces. 107 (2013) 97–106.
[20] P. Sathishkumar, M. Arulkumar, T. Palvannan, Utilization of agro-industrial waste Jatropha curcas
pods as an activated carbon for the adsorption of reactive dye Remazol Brilliant Blue R (RBBR), J.
Clean. Prod. 22 (2012) 67–75.
[21] M. Matouq, N. Jildeh, M. Qtaishat, M. Hindiyeh, M.Q. Al Syouf, The adsorption kinetics and
modeling for heavy metals removal from wastewater by Moringa pods, J. Environ. Chem. Eng. 3
(2015) 775–784.
[22] S.J. Allen, G. Mckay, J.F. Porter, Adsorption isotherm models for basic dye adsorption by peat in
single and binary component systems, J. Colloid Interface Sci. 280 (2004) 322–333.
15
[23] A. Mittal, J. Mittal, A. Malviya, D. Kaur, V.K. Gupta, Adsorption of hazardous dye crystal violet
from wastewater by waste materials, J. Colloid Interface Sci. 343 (2010) 463–473.
[24] M.B. Desta, Batch sorption experiments: Langmuir and freundlich isotherm studies for the
adsorption of textile metal ions onto teff straw (eragrostis tef) agricultural waste, J. Thermodyn.
1 (2013).
[25] K.Y. Foo, B.H. Hameed, Insights into the modeling of adsorption isotherm systems, Chem. Eng. J.
156 (2010) 2–10.
[26] W. Azlina, W. Ab, K. Ghani, A. Mohd, D.K. Mahmoud, N. Zalikha, Adsorption of methylene blue on
sawdust-derived biochar and its adsorption isotherms, J. Purity, Util. React. Environ. Vol.2 No.2,
April 2013, 34-50. 2 (2013) 34–50.
[27] O. Nakahara, Soil Science and Plant Nutrition Reconsideration of theoretical basis of Freundlich
adsorption isotherm equation Reconsideration of Theoretical Basis of Freundlich Adsorption
Isotherm Equation, Soil Sci. Plant Nutr. 768 (2016).
[28] L.E. Garcés Jaraba, S.C. Coavas Romero, Evaluación de la capacidad de adsorción en la cáscara de
naranja (Citrus SInensis) modificada con quitosano para la remoción de Cr(VI) en aguas
residuales, 2012. doi:10.1590/S0101-20612006000400017.
[29] D. Bulgariu, L. Bulgariu, Equilibrium and kinetics studies of heavy metal ions biosorption on green
algae waste biomass, Bioresour. Technol. 103 (2012) 489–493.
[30] M.T. Yagub, T.K. Sen, S. Afroze, H.M. Ang, Dye and its removal from aqueous solution by
adsorption: A review, Adv. Colloid Interface Sci. 209 (2014) 172–184.
[31] M. Xu, P. Hadi, G. Chen, G. McKay, Removal of cadmium ions from wastewater using innovative
electronic waste-derived material, J. Hazard. Mater. 273 (2014) 118–123.
[32] C. Moreno-Castilla, Adsorption of organic molecules from aqueous solutions on carbon materials,
Carbon N. Y. 42 (2004) 83–94.
[33] M.R. Mafra, D.R. Zuim, M.A. Ferreira, Adsorption of remazol brilliant blue on an orange peel
adsorbent, Brazilian J. 30 (2013) 657–665.
[34] S. Hashemian, J. Shayegan, A comparative studyof cellulose agricultural wastes (almond shell,
pistachio shell, walnut shell, tea waste and orange peel) for adsorption of violet B dye from
aqueous solutions, Orient. J. Chem. 30 (2014) 2091–2098.
[35] H. Gupta, B. Gupta, Adsorption of polycyclic aromatic hydrocarbons on banana peel activated
carbon, Desalin. Water Treat. (2015) 1–12.
[36] R. Popper, P. Niemz, S. Croptier, Adsorption and Desorption Measurements on Selected Exotic
Wood Species. Analysis with the Hailwood-Horrobin Model to Describe the Sorption Hysteresis,
Wood Res. 54 (2009) 43–55.
[37] N. Eggs, S. Salvarezza, R. Azario, Adsorción De Cromo Hexavalente En La Cáscara De Arroz
Modificada Químicamente, Av. En Ciencias E Ing. 3 (2012) 141–151.

16
Figure 1: Packed bed filtration system. (1) Feed water storage tank; (2) immersion pump; (3) feed control
valve; (4) column; (5) liquid level indicator; (6) fixed bed; (7) adjustable valve for level control; (8)
sample collection

100
90
80
Percentage Content , %

70
60
oTS
50
Ashes
40
Humidity
30
20
10
0
WS PS OP BN PP CB SD

Figure 2: Basic composition of the different adsorbents: organic material, ashes and water content

17
 a) 50 m 50 m

c) 50 m d) 50 m

e) 50 m

Figure 3: SEM images of raw samples. a) WS, b) PS, c) SD, d) OP, e) BP

18
Figure 4: Gas chromatogram of original produced water. Peaks: (1) Octadecanoic acid, (2) Benzoic acid,
(3) Hexadecanoic acid

Figure 5: Gas chromatogram of synthetic produced water. Peaks: (1) Tetramethyl, (2) oleic acid, (3)
octadecanoic acid

Figure 6: Gas chromatogram of SPW treated with WS. (1) Tetramethyl; (2) oleic acid; (3) octadecanoic
acid

19
 70
65
60
Removal, % 55
50
45
40
35
1 3 5 7 9
WS dosage, g

Figure 7: Percentage removal of oxidizable contaminants, represented as COD, according to WS dosage

(40 mL of SPW)

50
45
40
35
Removal, %

30
25
20
15
10
5
0
0 20 40 60 80 100 120
Contact time, min
Figure 8: Percentage removal of oxidizable contaminants represented as COD, according to the contact
time. (4 g of WS, 40 mL of SPW)

20
 60
1396 mg/L
50 780 mg/L
40 490 mg/L
Removal, % 377 mg/L
30

20

10

0
0 2 4 6 8 10
Contact time, min
Figure 9: Effect of the initial concentration on the percentage removal of oxidizable contaminants
represented as COD as function of time (4 g of WS, 40 mL of SPW)

60

50
Maximun Removal, %

WS
40
PS
30

20 SD

10

0
0 500 1000 1500
Initial COD concentration, mg/L

Figure 10: Comparison of the maximum removal of oxidizable contaminants represented as COD when
using WS, PS, SD

21
 a) b)
40 40
445 mg/L
35 35 780 mg/L
957 mg/L
30 30
1318 mg/L
Removal, %

Removal, %
25 25
20 20
15 368 mg/L 15
488 mg/L 10
10
788 mg/L
5 5
1404 mg/L
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Contact time, min Contact time, min

Figure 11: Effect of the initial concentration on the percentage removal of oxidizable contaminants
represented as COD using a) PS (4 g of PS, 40 mL of SPW) and b) SD (0.2 g of SD, 40 mL of SPW)

a) 4.5 b)
1396 mg/L 3.5
4.0 1404 mg/L
780 mg/L 3.0 788 mg/L
3.5
490 mg/L 2.5 488 mg/L
3.0 368 mg/L
qe, mg/g

377 mg/L
qe, mg/g

2.5 2.0
2.0 1.5
1.5
1.0
1.0
0.5 0.5

0.0 0.0
0 2 4 6 8 10 0 2 4 6 8 10
Contact time, min Contact time, min
c) 30

25

20
qe, mg/g

15
1318 mg/L
10 957 mg/L

5 780 mg/L
445 mg/L
0
0 2 4 6 8 10
Contact time, min

Figure 12: Adsorption capacity curves at equilibrium for different initial COD concentrations using a) WS
(4 g of WS, 40 mL of SPW), b) PS (4 g of PS, 40 mL of SPW), and c) SD (0.2 g of SD, 40 mL of SPW)

22
 a) 5.0 b) 5.0
Langmuir
4.0 4.0 Freundlich
Temkin
3.0 3.0 Experimental data
qe, mg/g

qe, mg/g
2.0 Langmuir 2.0
Freundlich
1.0 Temkin 1.0
Experimental data
0.0 0.0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Ce, mg/L Ce, mg/L

c) 35.0
30.0

25.0
qe, mg/g

20.0

15.0
Lagmuir
10.0 Freundlich
5.0 Temkin
Experimental data
0.0
0 500 1000 1500
Ce, mg/L

Figure 13: Nonlinear isotherms for a) WS, b) PS and c) SD

a) 1.6 b) 1.5
377 mg/L 368 mg/L
1.4 490 mg/L 1.0
488 mg/L
1.2 780 mg/L 0.5 788 mg/L
1296 mg/L 1404 mg/L
1.0 0.0
ln (qe-qt)
ln (qe-qt)

0.8 -0.5
0.6 -1.0
0.4 -1.5
0.2 -2.0
0.0 -2.5
0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0
Contact time, min Contact time, min

Figure 14: Linearized pseudo-first order kinetic model for a) WS b) PS

23
 1.00
0.95
WS
0.90
PS
0.85
C/C0 0.80
0.75
0.70
0.65
0.60
0 5 10 15 20 25 30
Time, min

Figure 15: Breakthrough curves for WS and PS

24