OLGA GOVEDARICA OPTIMIZATION OF THE ... No3_p357-368...Available on line at Association of the Chemical Engineers of Serbia AChE. Chem. Ind.Chem. Eng. Q. 24 (4) 357−368 (2018) CI&CEQ

Chemical Industry & Chemical Engineering Quarterly

Available on line at Association of the Chemical Engineers of Serbia AChE www.ache.org.rs/CICEQ

Chem. Ind. Chem. Eng. Q. 24 (4) 357−368 (2018) CI&CEQ

357

OLGA GOVEDARICA MILOVAN JANKOVIĆ

SNEŽANA SINADINOVIĆ-FIŠER

DRAGAN GOVEDARICA

Faculty of Technology, University of Novi Sad, Novi Sad, Serbia

SCIENTIFIC PAPER

UDC 665.345.4:66:544

OPTIMIZATION OF THE EPOXIDATION OF LINSEED OIL USING RESPONSE SURFACE METHODOLOGY

Article Highlights • Linseed oil was used as renewable raw material to produce epoxides • Epoxidation was conducted as batch process at approximately isothermal conditions • Response surface methodology was applied to optimize the epoxidation process • Relative epoxy yield was chosen as an objective function for the optimization • At the optimal conditions, product with high oxygen epoxy content was obtained Abstract

Epoxidized vegetable oils are widely used in the chemical industry. Their pro-duction requires the optimization of process conditions to maximize the epoxy yield. Therefore, the epoxidation of linseed oil with peracetic acid generated in situ in the presence of an ion exchange resin as a catalyst was optimized using response surface methodology combined with Box-Behnken design. The effects of temperature (65–85 °C), hydrogen peroxide-to-oil unsaturation mole ratio (1.1:1–1.5:1), catalyst amount (10–20 wt.%), and reaction time (5–13 h) on the epoxy yield were studied. According to analysis of variance, the developed regression model was significant with a coefficient of determination (R2) of 98.95%. Temperature of 70.6 °C, hydrogen peroxide-to-oil unsaturation mole ratio of 1.5:1, catalyst amount of 20 wt.%, and reaction time of 7 h were deter-mined as the optimal process conditions using the model. At these conditions, a relative epoxy yield of 84.73±0.07% was achieved, which agreed closely with the predicted value of 87.60%. The epoxidized linseed oil with high epoxy oxy-gen content (8.27±0.01%) and low iodine number (4.22±0.49 g iodine/100 g oil) was obtained approximately isothermally in a batch process and under rel-atively mild and safe conditions.

Keywords: Box-Behnken design, epoxidation, linseed oil, optimization, response surface methodology.

The use of epoxidized vegetable oils has seen increasing application diversity over the years. This non-toxic and biodegradable vegetable oil derivative is mainly used as a stabilizer and plasticizer for PVC and as an intermediate for the production of other vegetable oils derivatives [1,2]. It has also been iden-tified as a potential biolubricant [3,4]. The common oxidizing agent for the epoxidation of vegetable oils is

Correspondence: O. Govedarica, Faculty of Technology, Univer-sity of Novi Sad, Bul. cara Lazara 1, 21000 Novi Sad, Serbia. E-mail: [email protected] Paper received: 12 October, 2017 Paper revised: 23 January, 2018 Paper accepted: 3 April, 2018

https://doi.org/10.2298/CICEQ171012008G

percarboxylic acid, usually performic or peracetic acid. Percarboxylic acids are mainly generated in situ from the corresponding carboxylic acid and hydrogen peroxide in the presence of an acidic homogenous or heterogeneous catalyst. Percarboxylic acid spontane-ously reacts with the oil’s double bonds, forming an epoxy group and releasing carboxylic acid. Besides the main reactions of peroxidation and epoxidation, the acid-catalyzed side reactions of the epoxy group opening with nucleophilic agents also occur during the process [5-7]. As the epoxy yield and selectivity of the process are significantly influenced by the pro-cess conditions, their optimization is necessary reg-ardless of which carboxylic acid or catalyst is used in the peroxidation.

O. GOVEDARICA et al.: OPTIMIZATION OF THE EPOXIDATION OF LINSEED OIL… Chem. Ind. Chem. Eng. Q. 24 (4) 357−368 (2018)

358

For a long time, the classical “one-variable-at-a- -time” method has been used to investigate the effect of the process conditions on the epoxidation of vari-ous widely and locally available vegetable oils. It was applied in epoxidation studies of soybean [8,9], kar-anja [10,11], jatropha [12], mahua [13,14], rapeseed [15], castor [16], canola [17,18] and nahor [19] oil, as well as methyl esters derived from cottonseed oil [20]. The reaction temperature, mole ratios of carboxylic acid and hydrogen peroxide to vegetable oil unsatur-ation, catalyst amount, as well as the stirring speed were identified as the variables that have major effects on the epoxidation process [8–20].

More recently, experimental designs and statis-tical analyses have been employed to evaluate the effects of and to optimize the epoxidation process conditions. Response surface methodology, a statisti-cal method often used in chemical and biochemical engineering, was employed to optimize the epoxid-ation of vegetable oils such as rapeseed oil [21-23] and castor oil [24], as well as vegetable oil derivatives such as sucrose soyate [25] and methyl esters der-ived from castor oil [26], jatropha oil [27], and waste cooking oil [28]. Besides reducing the required expe-rimental work, this approach enables not only the det-ermination of the process conditions for reaching the highest epoxy yield, but also the evaluation of an interactive effect among the process conditions. Pre-vious studies on epoxidation have been conducted with peracetic acid generated in situ in the presence of either sulphuric acid [21,22] or the ion exchange resin Amberlite IR-120H [24–26,28] as the catalyst, with the exception of the epoxidation of rapeseed oil [23] and methyl esters derived from jatropha oil [27] which was conducted with performic acid. The effects of two [23], four [21,22,24,26–28], or five [25] process conditions including the reaction temperature, hydro-gen peroxide-to-oil unsaturation mole ratio, carboxylic acid-to-oil unsaturation mole ratio, catalyst amount, and reaction time, on the epoxy yield were analyzed in these studies.

The response surface methodology was com-bined either with the central composite [21-23,24,26– –28] or with the Box-Behnken [25] experimental design. Statistical analysis of the reported experi-mental data confirmed that all investigated conditions are significant for the epoxidation process in the applied ranges. Second-order polynomial equations were used in all studies for the prediction of either the relative epoxy yield [21,22,25,27] or the epoxy oxygen content [21–24,26,28]. In the developed models, i.e., regression equations, used for the calculation of the optimal process conditions, the linear and quadratic

terms were statistically significant, while in some of the studies either a few [27] or all [23,25] interaction terms were found insignificant.

The predicted optimal values of the process conditions differed among the studies, although, in some cases, the carboxylic acid, the concentration of hydrogen peroxide solution, as well as the type of the catalyst were the same and the investigated ranges of the process conditions were similar. Examples inc-lude the significant difference between the predicted optimal values of the reaction temperatures for the epoxidation of sucrose soyate [25] and castor oil [24]. For more unsaturated sucrose soyate, the optimal value of the reaction temperature (65 °C) was the highest one in the investigated range of 55-65 °C [25], while for castor oil the optimal value (52.81 °C) was close to the lowest temperature of the investigated range of 50-70 °C [24]. Also, the predicted optimal reaction time is 1.78 times longer for the epoxidation of more unsaturated sucrose soyate (5 h), even the optimal reaction temperature is higher than the one predicted for the epoxidation of castor oil. The com-parison of epoxy yields obtained under the optimal process conditions is not possible, since the yield has not been reported for the sucrose soyate.

Among the widely produced vegetable oils such as soybean oil, rapeseed oil, palm oil, or castor oil, linseed oil is the most unsaturated. This drying oil with mainly technical application is suitable for the pro-duction of epoxy derivative with high epoxy oxygen content. Such modified linseed oil can be further used to obtain thermosets of various properties by curing with different anhydrides [29,30] or dicarboxylic acids [31], or to produce waterborne coatings [32], as well as polyurethanes via non-isocyanate route by epoxy group opening with carbon dioxide [33]. Epoxy oxy-gen content, which determines the quality and the application of epoxidized linseed oil, depends on the applied process conditions for the epoxidation. The effects of the process conditions and their interactions on the formation of the epoxy groups have not been previously determined for the epoxidation of linseed oil with peracetic acid. Therefore, the aim of the pre-sent work was to evaluate the effects of and to opti-mize the process conditions by maximizing the epoxy yield for the epoxidation of linseed oil with peracetic acid generated in situ in the presence of an ion exchange resin as the catalyst. The effects of four process conditions (reaction temperature, hydrogen peroxide-to-linseed oil unsaturation mole ratio, catal-yst amount, and reaction time) on the epoxy yield were studied by employing the Box-Behnken experimental design within the response surface methodology.


359

EXPERIMENTAL

Materials

Linseed oil was bought from a local market. The hydrogen form of acidic ion exchange resin Amberlite IR-120H was purchased from Sigma-Aldrich (St. Louis, MO, USA). Hydrogen peroxide (30% aqueous solution), iodine (p.a.), and bromine (p.a.) were bought from Centrohem (Stara Pazova, Serbia). Gla-cial acetic acid, potassium iodide (extra pure), chloro-form (min 98.5%), chlorobenzene (pure), and potas-sium hydrogen phthalate (min 99.0%) were pur-chased from LachNer (Neratovice, Czech Republic). Sigma-Aldrich (St. Louis, MO, USA) was the supplier of hydrogen bromide solution (33.0%) in acetic acid and crystal violet. Aqueous solution of sodium thio-sulfate (0.1 N) was bought from Alfapanon (Novi Sad, Serbia).

Epoxidation procedure

The linseed oil was epoxidized with peracetic acid generated in situ through the reaction of glacial acetic acid and 30% aqueous solution of hydrogen peroxide in the presence of Amberlite IR-120H ion exchange resin as the catalyst, according to the modi-fied method reported in the literature [8,9]. The lin-seed oil was characterized with an initial iodine num-ber of 171.5 g iodine/100 g oil, which corresponds to the unsaturation level of 0.6757 mol of double bond per 100 g of oil. The epoxidation runs were carried out at temperatures of 65, 75 and 85 °C. The applied hydrogen peroxide-to-oil unsaturation mole ratios were 1.1:1, 1.3:1 and 1.5:1, while the mole ratio of acetic acid-to-oil unsaturation was 0.5:1 in all runs. Linseed oil (approximately 80 g), acetic acid and hyd-rogen peroxide were loaded into a 500 mL three-neck round bottom glass reactor. The reactor, equipped with a thermometer, a condenser, and PTFE-coated cylindrical magnetic stirring bar (35 mm×6 mm), was immersed in a temperature-controlled water bath. When the stirred mixture reached a desired tempe-rature, the catalyst was added into the reactor at once. This point was considered to be the “zero time” of the process. The used catalyst amounts, expres-sed as a percentage of acetic acid and 30% hydrogen peroxide weight, were 10, 15 and 20 wt.%. The fine dispersion of the linseed oil in the reaction mixture was achieved with the uniform agitation under the constant stirring speeds of 1600, 1800 and 2000 rpm. The temperature of the reaction mixture was main-tained at the desired value with the fluctuation of less than ±1 °C. During the run, when needed, the tempe-rature of the reaction mixture was adjusted by cooling

or heating the water bath. The runs lasted for 5, 9 and 13 h. At the end of the run, the reaction mixture was cooled and centrifuged at least 5 min under the rot-ation speed of 2000 rpm. The oil phase, separated from the aqueous phase and the catalyst, was then washed with water (about 50 °C) until acid free, and evaporated for at least 1 h under the vacuum of about 30 mbar at 60 °C. The oil phase samples were further analyzed to determine the iodine number and the epoxy oxygen content.

Characterization of the linseed oil, epoxidized linseed oil and the epoxidation process

The iodine number of the linseed oil and the oil phase samples was determined according to the Hanus method [34]. AOCS Cd 9-57 method was applied to determine the epoxy oxygen content of the oil phase samples [35]. All analyses were performed in triplicate.

The theoretical maximal content of epoxy oxy-gen of linseed oil (EOt, %) is calculated according to the following expression:

=

0

It O

0O

I

1002

100+2

INAEO AIN A

A

(1)

where IN0 is the initial iodine number of linseed oil and AO and AI are the atomic masses of oxygen and iodine, respectively. The relative epoxy yield (Y, %) is defined as follows:

=t

100EOYEO

(2)

where EO (%) is the epoxy oxygen content of the oil phase sample. The iodine number of the oil phase sample (IN) was used to calculate the conversion of double bonds (X, %) according to the following expression:

−= 00

100IN INX

IN (3)

The selectivity of the epoxidation process (SE) is defined as:

= YSEX

(4)

Characterization of the catalyst

The amount of sulpho groups in the ion exchange resin Amberlite IR-120H was determined by the titration method. To achieve ion exchange between H+ and Na+, the resin suspension in NaCl


360

solution (200 g/L) was stirred for 24 h at the room temperature. Afterwards, the solution was titrated with 0.1N NaOH solution [36,37].

Experimental design

A Box-Behnken design of experiments was employed in the optimization study of the epoxidation of linseed oil with peracetic acid formed in situ in the presence of the ion exchange resin Amberlite IR- -120H. The reaction temperature (X1), hydrogen per-oxide-to-oil unsaturation mole ratio (X2), catalyst amount (X3) and reaction time (X4) were chosen as four independent variables, i.e., factors, whose indi-vidual and interaction effects on the relative epoxy yield (Y) as the dependent variable, i.e., response, were evaluated. The actual and coded values of the factors, the latter values corresponding to the low (-1),

middle (0), and high (+1) levels, are presented in Table 1. The experimental matrix of the three-level four-factor Box-Behnken design, consisting of 24 experimental points and 3 replicates of the central point, together with the results is given in Table 2.

Table 1. Process factors and their uncoded (actual) and coded levels

Factor Level

-1 0 +1

Temperature, X1 (°C) 65 75 85 Hydrogen peroxide-to-oil unsaturation mole ratio, X2

1.1 1.3 1.5

Catalyst amount, X3 (wt.%) 10 15 20 Reaction time, X4 (h) 5 9 13

Table 2. Experimental matrix of the Box-Behnken design with the experimentally determined (exp.) and calculated (calcd.) relative epoxy yields. The relative deviation of the results is presented; catalyst amount (X3) is expressed as the percentage of acetic acid and hydrogen peroxide aqueous solution weights. Values in parenthesis represent the catalyst amount expressed as the percentage of the linseed oil weight

Run Coded factor Uncoded factor Relative epoxy yield, Y / %

X1 X2 X3 X4 X1 / °C X2 X3 / wt.% X4 / h Exp. Calcd. Relative deviation, RD / %

1 0 1 -1 0 75 1.5 10 [13.5] 9 84.61±0.03 85.16 -0.65

2 -1 -1 0 0 65 1.1 15 [15.6] 9 75.23±0.17 75.47 -0.32

3 -1 0 0 -1 65 1.3 15 [18.0] 5 69.27±0.61 69.16 0.16

4 0 0 -1 -1 75 1.3 10 [12.0] 5 71.01±0.75 71.96 -1.3

5 -1 0 -1 0 65 1.3 10 [12.0] 9 73.47±0.08 74.47 -1.4

6 1 0 0 -1 85 1.3 15 [18.0] 5 66.34±0.93 66.52 -0.27

7 0 0 0 0 75 1.3 15 [18.0] 9 84.11±0.41 83.14 1.2

8 0 1 0 -1 75 1.5 15 [20.3] 5 82.55±0.07 82.42 0.16

9 1 0 0 1 85 1.3 15 [18.0] 13 56.86±0.16 56.83 0.053

10 0 0 -1 1 75 1.3 10 [12.0] 13 83.15±0.95 83.29 -0.17

11 1 -1 0 0 85 1.1 15 [15.6] 9 63.39±0.13 64.73 -2.1

12 0 1 0 1 75 1.5 15 [20.3] 13 78.46±0.33 79.33 -1.1

13 0 -1 1 0 75 1.1 20 [20.9] 9 83.30±0.50 82.62 0.82

14 0 -1 0 1 75 1.1 15 [15.6] 13 78.57±0.07 79.66 -1.4

15 0 0 0 0 75 1.3 15 [18.0] 9 81.80±0.18 83.14 -1.64

16 -1 0 1 0 65 1.3 20 [24.0] 9 81.97±0.75 83.32 -1.65

17 0 1 1 0 75 1.5 20 [27.0] 9 82.11±0.34 83.35 -1.5

18 1 0 1 0 85 1.3 20 [24.0] 9 62.40±0.30 62.36 0.064

19 0 0 1 1 75 1.3 20 [24.0] 13 76.21±0.60 75.19 1.34

20 -1 1 0 0 65 1.5 15 [20.3] 9 84.26±0.42 82.84 1.68

21 0 -1 0 -1 75 1.1 15 [15.6] 5 73.47±0.09 73.56 -0.12

22 0 -1 -1 0 75 1.1 10 [10.5] 9 78.73±0.11 77.36 1.7

23 0 0 0 0 75 1.3 15 [18.0] 9 83.14±0.38 83.14 0

24 0 0 1 -1 75 1.3 20 [24.0] 5 83.72±0.45 83.51 0.25

25 1 0 -1 0 85 1.3 10 [12.0] 9 68.15±0.15 67.75 0.59

26 1 1 0 0 85 1.5 15 [20.3] 9 66.21±0.05 65.89 0.48

27 -1 0 0 1 65 1.3 15 [18.0] 13 82.17±0.44 81.86 0.38

Mean relative percentage deviation, MRPD / % ±0.83


361

Statistical analysis

To determine the relationship between the relat-ive epoxy yield and the epoxidation process condi-tions, multiple nonlinear regression of the experimen-tal data was performed using Statistica 12.0 software package (StatSoft, USA), in order to develop a sec-ond-order polynomial model:

α α α α= = <

= + + + 201 1

n n n

j j jj j kj j kj j k j

Y X X X X (5)

where Y is the response, i.e., dependent variable (rel-ative epoxy yield); Xj and Xk are the factors, i.e., inde-pendent variables; n indicates the number of indepen-dent variables; α0 is the offset term; and αj, αjj and αkj are the linear, squared, and interaction effect terms, respectively. The quality-of-fit of the model was exp-ressed by the coefficient of determination (R2) and the mean relative percentage deviation (MRPD), which is defined as:

==

1

NR

ii

RDMRPD

NR (6)

where RDi (%) is the relative deviation of the relative epoxy yield determined for run i; and NR is the total number of runs. RDi is calculated as:

−=exp calc

exp100i i

ii

Y YRDY

(7)

where superscripts exp and calc denote experiment-ally determined and calculated value of the relative epoxy yield, respectively.

The analysis of variance (ANOVA) was used to evaluate the significance of the lack-of-fit of the model, as well as the significance of the terms in the regression model at the confidence level of 95%.

RESULTS AND DISCUSSION

Preliminary experiments and ranges of the process conditions

Linseed oil, as a highly unsaturated vegetable oil, is a valuable raw material for chemical transform-ations such as epoxidation, which occur at the double bond sites of triglycerides. The determined value of the initial iodine number (171.5 g iodine/100 g oil) corresponds to the theoretical maximal content of epoxy oxygen of 9.76%, calculated using Eq. (1).

The epoxidation of vegetable oils with percar-boxylic acid generated in situ is usually performed as a semi-batch process with drop-wise or pulse-wise addition of the hydrogen peroxide solution [7–28].

Such gradual addition of the hydrogen peroxide is applied in order to control the exothermic effect of the percarboxylic acid formation. However, with efficient heat removal, a single addition of the hydrogen per-oxide solution is possible, enabling faster completion of the epoxidation process [38]. Under the experimen-tal conditions applied in this work, i.e., 30% hydrogen peroxide solution and the adjustment of the water bath temperature, approximately isothermal condit-ions were achieved during the batch epoxidation runs.

To determine the conditions for efficient external mass transfer in the liquid-liquid-solid (oil-aqueous- -resin) reaction system of linseed oil epoxidation in the presence of Amberlite IR-120H, the influence of the stirring speed on the conversion of double bonds and relative epoxy yield was investigated. The epo-xidation runs were performed at stirring speeds rang-ing between 1600-2000 rpm, at the reaction condit-ions corresponding to the central point of the emp-loyed Box-Behnken experimental design (Table 1). The conversion of double bonds and the relative epoxy yield (Figure 1) did not change significantly with an increase in stirring speed above 1800 rpm. To ensure the kinetic regime, i.e., reaction rates signific-antly higher than the mass diffusion rates, all epoxid-ation runs included in the Box-Behnken experimental matrix were performed at a stirring speed of 2000 rpm.

Figure 1. Conversions of double bonds and relative epoxy yields

achieved after 9 h of the epoxidations of linseed oil in the presence of 15 wt.% of Amberlite IR-120H at 75 °C, at oil

unsaturation-to-acetic acid-to-hydrogen peroxide mole ratio of 1:0.5:1.3 and stirring speeds of 1600, 1800 and 2000 rpm.

The acetic acid undergoes perhydrolysis in the reaction system of the epoxidation, whereas it is rel-eased when the vegetable oil’s double bonds are epoxidized by the peracetic acid. This carboxylic acid is one of the nucleophilic agents that open the epoxy group in the presence of the acidic ion exchange


362

resin [6]. Therefore, due to its “recycle” and negative influence on the process selectivity, the acetic acid is usually applied in stoichiometric deficiency with res-pect to both hydrogen peroxide and double bond amounts. The common mole ratio of acetic acid-to- -double bond of 0.5:1, used in other studies [8–14,16– -19,24,26,28], was also applied in this work for all lin-seed oil epoxidations.

The hydrogen peroxide is usually applied in excess with respect to acetic acid amount and also to vegetable oil’s double bond amount. The higher amount of hydrogen peroxide in the reaction system increases the reaction rate of the peracetic acid form-ation. The faster formation of the peracetic acid leads to the higher reaction rate of the double bond epoxid-ation with peracetic acid. Therefore, the higher amount of hydrogen peroxide shortens the reaction time nec-essary for reaching the maximum epoxy yield. Due to the shorter reaction time of the epoxidation process, there is less time for the side reactions of epoxy group opening to occur. However, a higher amount of hyd-rogen peroxide in the reaction system may have a negative impact on the selectivity of the epoxidation process and on the obtained maximal epoxy yield since hydrogen peroxide is the nucleophilic agent that opens the epoxy group [6]. Several studies confirmed that the maximal epoxy yield increases with an inc-rease in excess of hydrogen peroxide up to 50%. Above the excess of 50%, the obtained results dif-fered among the studies: in some of them the maxi-mal epoxy yield decreased [14,28], whereas in the others it stayed almost constant [11,12] or even inc-reased [18]. Hydrogen peroxide is applied as an aqueous solution, therefore higher amounts of hydro-gen peroxide increases the volume of the waste water generated after the epoxidation process. Due to this negative environmental impact, the excess of hydro-gen peroxide higher than 50% with respect to the double bond amount was not applied in this study (Table 1).

The catalyst Amberlite IR-120H is the cation exchange resin of the sulpho type. The concentration of the active sites, i.e., sulpho groups, in the catalyst used in this work was determined as 2.4 meq per 1 g of wet resin. The amount of Amberlite IR-120H was 10, 15, or 20 wt.% with respect to the total mass of acetic acid and aqueous solution of hydrogen per-oxide applied in the run (Table 1).

The decrease of the reaction temperature inc-reases the selectivity of the process, but prolongs the reaction time necessary to reach the maximum epoxy yield [11,14,16]. In the preliminary experiments (con-ducted at 2000 rpm, 60 °C, 20 wt.% catalyst and

hydrogen peroxide-to-acetic acid-oil unsaturation mole ratio 1.5:0.5:1), the maximal relative epoxy yield of 85.52±0.43%, with complete conversion of double bonds, was reached after 17 h. To avoid such long reaction times, which are not of interest for the ind-ustry, the effect of temperature on the epoxidation of linseed oil was investigated by applying 65 °C as the lowest value (Table 1). Due to safety concerns, the effect of temperature was not investigated above 85 °C.

Regression of the experimental data

For the optimization of the epoxidation of veget-able oils, central composite design has most often been employed within the response surface method-ology [21–24,26–28]. The axial experimental points of this experimental design are usually located beyond the investigated ranges of the process conditions. Hence, for the temperature range investigated in this work (65-85 °C), the epoxidation runs would also have to be conducted at temperatures higher than 85 °C. To avoid such high temperatures, the Box-Behn-ken design was employed, since its experimental points are inside the investigated ranges of the pro-cess condition values.

The relative epoxy yields obtained for the epo-xidation runs performed according to the experimental matrix of the Box-Behnken design are presented in Table 2. The values of the relative epoxy yield ranged from 56.86±0.16 to 84.61±0.03% for all runs. Using multiple regression of the experimental data, the fol-lowing model equation with uncoded factors was obtained:

= − + + + +

+ − − − −

− − − −− − − −−

1 2 3

2 2 24 1 2 3

24 1 2 1 3

1 4 2 3 2 4

3 4

869.1 18.44 146.0 10.79

22.62 0.1053 9.521 0.02553

0.2512 0.7763 0.071250.1399 1.767 2.8720.2456

Y X X X

X X X X

X X X X XX X X X X XX X

(8)

The good quality-of-fit of the experimental data to the regression model is confirmed by the high value of the coefficient of determination (R2) of 98.95% indicating that only 1.05% of the variation in the experimental data cannot be explained by the model. The comparison between the experimentally determined and predicted relative epoxy yields is pre-sented in Figure 2. The high extent of agreement between the experimentally determined and calcul-ated values of the relative epoxy yield is also con-firmed by the low value of the mean relative percent-age deviation (MRPD) of 0.83% (Table 2). The good predictability of the developed regression model is implied by reasonable agreement between the pre-


363

dicted coefficient of determination (94.53%) and the adjusted coefficient of determination (97.73%). The test of significance showed that the model can be considered as highly significant (p < 0.0001), whereas lack-of-fit of the model is insignificant (p > 0.05) at the confidence level of 95% in the investigated ranges of the process conditions. The good quality-of-fit of the experimental data and the predictability of the regres-sion equation were also obtained in the other optimiz-ation studies of the vegetable oil epoxidations with percarboxylic acid, regardless of the applied experi-mental design [21–28].

Figure 2. Scatter plot of the predicted and experimentally

determined values of the relative epoxy yield (Y) for all runs included in the Box-Behnken experimental matrix.

All four process conditions (temperature, hydro-gen peroxide-to-oil unsaturation mole ratio, catalyst amount, and reaction time) positively influence the relative epoxy yield since regression coefficients of their linear terms have positive values. The negative values of all quadratic terms indicate that the regres-sion model is concave.

The significance of the process conditions and related terms in the regression model is assessed using Student’s test. The calculated t-values of terms are presented on the Pareto chart in Figure 3.

The highest t-values of linear and quadratic terms of the process factor X1 indicate that the tem-perature has the most significant influence on the rel-ative epoxy yield in the applied ranges of process conditions. This finding is in agreement with two rep-orted optimization studies in which the influence of the same four factors on the epoxidation was inves-tigated. Those were the optimization studies of the epoxidation of castor oil [24] and castor oil methyl esters [26] with peracetic acid formed in situ in the presence of the ion exchange resin. Since the react-ion rates significantly change with an increase in reaction temperature, the most significant influence of

this factor may be explained by the wide range of its applied values of 20 °C in the present, as well as in the reported studies.

Figure 3. Pareto chart of t-values of all terms in the regression

model.

All terms in the regression model, Eq. (8), whose p-values are given in Table 3, were revealed to be significant (p < 0.05), except the square terms of the hydrogen peroxide-to-oil unsaturation mole ratio (X22) and the catalyst amount (X32). The values of variance of the inflation factor of the terms in the reg-ression model (VIF < 5) confirmed that there is no sig-nificant correlation between the independent variables.

Table 3. ANOVA results for the developed regression model

Source p-Value VIF

Model


364

Influence of the process conditions on the relative epoxy yield

The contour plots obtained based on the reg-ression model, Eq. (8), shown in Figures 4–6, are used to investigate the dependence of the relative epoxy yield on the two process conditions, while the other two are at the middle values applied in the Box- -Behnken design (Table 1).

The hydrogen peroxide-to-oil unsaturation mole ratio significantly influences the relative epoxy yield at the temperatures below 77 °C (Figure 4a). In the

temperature range from 69 to 73 °C, the relative epoxy yield constantly increases from 81 to above 85% with an increase in the mole ratio from 1.1 to 1.5. This is due to the increase in the reaction rate of the peracetic acid formation, which consecutively inc-reases the rate of the epoxidation. In this case, the reaction time for the occurrence of the side reactions is shortened and the selectivity of the epoxidation process is increased. For instance, the relative epoxy yield of 83.15±0.95% was achieved with the double bond conversion of 98.45±0.03% and the selectivity of 0.8446 at 75 °C in the presence of 10 wt.% of

Figure 4. Contour plots of the relative epoxy yield (Y) as a function of: a) temperature and hydrogen peroxide-to-oil unsaturation mole

ratio and b) temperature and catalyst amount.

Figure 5. Contour plots of the relative epoxy yield (Y) as a function of: a) temperature and reaction time and b) hydrogen peroxide-to-oil

unsaturation mole ratio and catalyst amount.

Figure 6. Contour plots of the relative epoxy yield (Y) as a function of: a) hydrogen peroxide-to-oil unsaturation mole ratio and reaction

time and b) catalyst amount and reaction time.


365

Amberlite IR-120H after 13 h when the hydrogen per-oxide-to-oil unsaturation mole ratio was 1.3 (run 10, Table 2). With higher hydrogen peroxide-to-oil unsat-uration mole ratio of 1.5, higher relative epoxy yield of 84.61±0.03% was achieved, with higher double bond conversion of 99.15±0.13% and higher selectivity of the epoxidation process of 0.8534 after only 9 h (run 1, Table 2). The relative epoxy yields are the highest (above 85%) for central values of the catalyst amount and reaction time when the hydrogen peroxide-to-oil unsaturation mole ratio is higher than 1.45 and at temperature close to 71 °C (Figure 4a).

The influence of the catalyst amount on the rel-ative epoxy yield also depends on the temperature value (Figure 4b). At temperatures higher than 76 °C, the influence of the catalyst on the relative epoxy yield is less significant. The simultaneous influence of high temperatures and high catalyst amounts leads to the lowest values of the relative epoxy yield, because high temperatures promote the side reactions result-ing in the lower stability of the epoxy group [11,14]. Likewise, the higher amount of the catalyst increases the rates of the epoxy group opening reactions, which are catalyzed by the acidic active sites on the external surface of the resin [6]. In the temperature range from 67 to 73 °C, the relative epoxy yield increases from 80 to above 85% with the increase in the applied catalyst amount from 10 to 20 wt.% with respect to the total mass of acetic acid and hydrogen peroxide aqueous solution. The highest relative epoxy yields above 85% are achieved at the highest applied amounts of the catalyst (above 17 wt.%) and lower values of the temperature (around 67–73 °C) in the investigated ranges of process conditions.

Since temperature is a kinetic factor, relative epoxy yield highly depends on the simultaneous influ-ence of the temperature and the reaction time (Figure 5a). Lower temperatures and longer reaction times promote the highest relative epoxy yields in the applied ranges of the process conditions. Goud et al. [11] found that the optimal temperatures are around 70 °C for the epoxidation of karanja oil with peracetic acid formed in situ in the presence of Amberlite IR- -120H. The highest relative epoxy yield of 85% was achieved after 4 h at this temperature. However, for the epoxidation of linseed oil the reaction times that correspond to the highest relative epoxy yields (above 80%) is longer (9–12.5 h) at the temperatures around 70 °C (Figure 5a) due to almost two-fold higher unsat-uration of linseed oil (initial iodine number 171.5 g iodine/100 g oil) compared to karanja oil (initial iodine number 89 g iodine/100 g oil).

The interaction effect of the catalyst amounts and the hydrogen peroxide-to-oil unsaturation mole ratios on the relative epoxy yield can be discussed on the basis of the contour plot presented in Figure 5b. When the hydrogen peroxide-to-oil unsaturation mole ratio is lower than 1.4, the influence of the catalyst amount on the relative epoxy yield is highly signific-ant. Thus, the relative epoxy yield increases with an increase of the catalyst amount (from 10 to 20 wt.%). Above the hydrogen peroxide-to-oil unsaturation mole ratio of around 1.45, the relative epoxy yield dec-reases gradually below 84% with the increase in the catalyst amount above 18 wt.%.

For the reaction time range from 5 to 12 h an increase of the hydrogen peroxide-to-oil unsaturation mole ratio in the applied range from 1.1 to 1.5 inc-reases the value of the relative epoxy yield (Figure 6a). This is in agreement with findings in several pub-lished investigations of the process conditions influ-ence on the epoxidation of different vegetable oils in which the “one-variable-at-a-time” method was applied [11,12,14,18].

Based on the contour plot given in Figure 6b, it is obvious that the influence of the catalyst amount on the relative epoxy yield is highly dependent on the reaction time. The highest values of the relative epoxy yield (above 84%) are achieved in the presence of more than 19 wt.% of the catalyst after reaction time of 6 to 8 h. The relative epoxy yields lower than 76% are obtained for higher catalyst amounts (above 19 wt.%) and for reaction times longer than 12.5 h. High amounts of the catalyst Amberlite IR-120H, which also catalyzes epoxy group opening reactions with nucleophilic agents, cause high rates of these side reactions. At the same time, due to longer reaction times, the amount of epoxy group is high in the react-ion system. Therefore, high amounts of the catalyst and longer reaction times have a favorable effect on the rates of the epoxy group opening reactions. The mentioned combination of the process conditions cor-responds to the process period when the relative epoxy yield already reached the maximum value and it is now decreasing due to the high rates of the side reactions. Relative epoxy yields lower than 76% are also obtained for the lower catalyst amounts (below 13 wt.%) and shorter reaction times (below 6.5 h). This combination of the process values corresponds to the process period when the relative epoxy yield is still increasing up to the maximum.

Validation of the regression model

The regression model, Eq. (8), was developed in order to determine the optimal conditions for reaching


366

maximal relative epoxy yield in the applied process conditions ranges for the epoxidation of linseed oil with peracetic acid formed in situ in the presence of the ion exchange resin Amberlite IR-120H. The model was validated on the basis of three additional runs by comparing the experimentally determined and pre-dicted values of the relative epoxy yield. The sets of the process conditions for additional runs differed from all runs included in the Box-Behnken experi-mental matrix. The process conditions and obtained relative epoxy yields of the validation experiments V1, V2, and V3 are presented in Table 4. The relative deviation between the experimentally determined and predicted values of the relative epoxy yield for these experiments ranged from -1.8 to 6%. The highest absolute value of the relative deviation of 6% was obtained for run V3, conducted under three boundary values of the applied process conditions ranges, i.e., maximal temperature of 85 °C, maximal hydrogen peroxide-to-oil unsaturation mole ratio of 1.5, and minimal catalyst amount of 10 wt.%. The lowest absolute value of the relative deviation of 0.15% was obtained for run V2 when none of the process con-ditions had boundary values, either maximum or mini-mum. Although the mean percentage relative devi-ation for the validation runs of ±2.7% (Table 4) is 3.3 times higher than the mean percentage relative devi-ation for the runs included in the Box-Behnken matrix (Table 2), this value indicates that the regression model satisfactorily predicts the relative epoxy yield in the applied ranges of the process conditions.

In order to compare the value of the mean percentage relative deviation obtained in this work with those obtained in the similar studies, the devi-ation was calculated for the validation experiments performed in the optimization study of the sucrose soyate epoxidation with peracetic acid formed in situ in the presence of the ion exchange resin [25]. The calculated mean relative percentage deviation for seven reported validation experiments was ±0.72%. The lower value of the mean relative percentage deviation in the reported study can be explained by the smaller share of the experiments conducted under the boundary values of the process conditions than in

the present study. Only three of seven validation exp-eriments in the study of the sucrose soyate epoxid-ation were conducted under the process conditions with boundary values. Moreover, in these three expe-riments only one of five process conditions had the boundary value.

Determination of the optimal process conditions

The highest value of the relative epoxy yield of 87.60% is predicted by the developed regression model in the applied ranges of the process condition for the following set of the values: temperature of 70.6 °C, hydrogen peroxide-to-oil unsaturation mole ration of 1.5, catalyst amount of 20 wt.%, and reaction time of 7 h. In the control experiment, conducted under these conditions, the relative epoxy yield of 84.73± ±0.07% was achieved. The yield is obtained with double bond conversion of 97.54±0.28% and select-ivity of the epoxidation process of 0.8687. Although the experimentally determined relative epoxy yield is 3.28% lower than the predicted value, it is higher than any of the relative epoxy yields obtained in the runs included in the Box-Behnken experimental matrix (Table 2). Therefore, the calculated values of the pro-cess conditions can be considered as the optimal values for the epoxidation of linseed oil with peracetic acid formed in situ in the presence of Amberlite IR-120H.

The optimal process conditions were also rep-orted for the epoxidation of another vegetable oil, namely castor oil, with peracetic acid formed in situ in the presence of the ion exchange resin [24]. The same four process conditions were optimized as in the present study, but the applied ranges of their values were different. Temperature of 52.8 °C, hydro-gen peroxide-to-oil unsaturation mole ration of 1.65:1, catalyst amount of 15.14 wt.%, and reaction time of 2.81 h were determined as the optimal process conditions for the epoxidation of castor oil. The optimization was performed in order to maximize the epoxy oxygen content. Since the iodine number of the castor oil was given, it is possible to calculate the cor-responding relative epoxy yields using Eqs. (1) and (2). The predicted value of the relative epoxy yield was 76.43%, whereas the experimentally determined

Table 4. Experimental conditions used for the validation of the regression model and the results obtained experimentally (exp.) and calculated (calcd.) by applying the regression model

Run X1 / °C X2 X3 / wt.% X4 / h Relative epoxy yield, Y / %

Exp. Calcd. RD / %

V1 75 1.1 10 11 78.94 ±0.37 80.34 -1.8

V2 75 1.3 15 11 82.39 ±0.28 82.51 -0.15

V3 85 1.5 10 7 82.04 ±0.61 69.83 6.0

Mean relative percentage deviation, MRPD / % ±2.7


367

relative epoxy yield was 71.94%. Both values are lower than the corresponding values obtained in the present optimization study of the linseed oil epoxid-ation.

CONCLUSIONS

The Box-Behnken experimental design and res-ponse surface methodology were used for modeling and optimization of the epoxidation of linseed oil with peracetic acid generated in situ from acetic acid and hydrogen peroxide in the presence of the hetero-geneous catalyst. The developed regression model satisfactorily predicted the relative epoxy yield for new combinations of the process conditions with low mean relative percentage deviation of ±2.7%. The relative epoxy yield of 84.73±0.07% and almost complete conversion of linseed oil double bonds were achieved after 7 h at a temperature of 70.6 °C, hydrogen per-oxide-to-oil unsaturation mole ratio of 1.5:1 and cat-alyst amount of 20 wt.%. Under these optimal condi-tions, the process yielded epoxidized linseed oil with epoxy oxygen content of 8.27±0.01%. Such high epoxy oxygen content is a desirable property for the application of this oil derivative in the chemical and polymer industry.

Acknowledgement

This work was supported by the Ministry of Edu-cation, Science and Technological Development of the Republic of Serbia (Project #III45022).

Nomenclature

A Atomic mass EO Epoxy oxygen content (%) EOt Theoretical maximal content of epoxy oxygen

(%) IN Iodine number (g iodine/100 g oil) IN0 Initial iodine number (g iodine/100 g oil) MRPD Mean relative percentage deviation (%) RD Relative deviation (%) SE Selectivity of the epoxidation process VIF Variance of inflation factor X Conversion of double bonds (%) X1 Temperature (°C) X2 Hydrogen peroxide-to-oil unsaturation mole

ratio X3 Catalyst amount (wt.%) X4 Reaction time (h) Y Relative epoxy yield (%)

Superscript exp Experimentally determined calc Calculated

REFERENCES

[1] L.A.Rios, P.P.Weckes, H. Schuster, W.F. Hoelderich, Appl. Catal., A 284 (2005) 155–161

[2] D.D.Plaza, V.Strabel, P.K.K.S.Heer, A.B.Sellars, S-S.Hoong, A.Clark, A.Lapkin, J. Chem. Technol. Biotech-nol. 92 (2017) 2254–2266

[3] X. Wu, X. Zhang, S. Yang, H. Chen, D. Wang, J. Am. Oil Chem. Soc. 77 (2000) 561–563

[4] A. Adhvaryu, S.Z. Erhan, Ind. Crops Prod. 15 (2002) 247– –254

[5] A. Campanella, M.A. Baltanás, Chem. Eng. J. 118 (2006) 141–152

[6] A. Campanella, M.A. Baltanás,Chem. Eng. Process. 46 (2007) 210–221

[7] Z. Wu, Y. Nie, W. Chen, L. Wu, P. Chen, M. Lu, F. Yu, J.Ji, Can. J. Chem. Eng. 94 (2016) 1576–1582

[8] S. Sinadinović-Fišer, M. Janković, Z. Petrović, J. Am. Oil Chem. Soc. 78 (2001) 725–731

[9] M. Janković, S. Sinadinović-Fišer, O. Govedarica, J. Pav-ličević, J. Budinski-Simandić, Chem. Ind. Chem. Eng. Q. 23 (2017) 97–111

[10] V.V. Goud, N.C. Pradhan, A.V. Patwardhan, J. Am. Oil Chem. Soc. 83 (2006) 635–640

[11] V.V. Goud, A.V. Patwardhan, S. Dinda, N.C. Pradhan, Eur. J. Lipid Sci. Technol. 109 (2007) 575–584

[12] V.V. Goud, A.V. Patwardhan, S. Dinda, N.C. Pradhan, Chem. Eng. Sci. 62 (2007) 4065–4076

[13] V.V. Goud, A.V. Patwardhan, N.C. Pradhan, Bioresour. Technol. 97 (2006) 1365–1371

[14] V.V. Goud, A.V. Patwardhan, N.C. Pradhan, Ind. Eng. Chem. Res. 46 (2007) 3078–3085

[15] E. Milchert, A. Smagowicz, J. Am. Oil Chem. Soc. 86 (2009) 1227–1233

[16] S. Sinadinović-Fišer, M. Janković, O. Borota, Chem. Eng. Process. 62 (2012) 106–113

[17] E. Monono, D. Hagenson, D. Wiesenborn, Ind. Crops Prod. 67 (2015) 364–372

[18] R. Mungroo, V. Goud, S.N. Naik, A. Dalai, Eur. J. Lipid Sci. Technol. 113 (2011) 768–774

[19] S. Dinda, V. Ravisankar, P. Puri, J. Taiwan Inst. Chem. Eng. 65 (2016) 399–404

[20] X. Gui, Y. Ding, Z. Yun, Can. J. Chem. Eng. 94 (2016) 424–429

[21] E. Milchert, A. Smagowicz, G. Lewandowski, J. Chem. Technol. Biotechnol. 85 (2010) 1099–1107

[22] E. Milchert, A. Smagowicz, G. Lewandowski, Org. Pro-cess. Res. Dev.14 (2010) 1094–1101

[23] S. Arumugam, G. Sriram, T. Rajmohan, Arab. J. Sci. Eng. 39 (2014) 2277-2287

[24] V.B. Borugadda, V. Goud, Energy Sci. Eng. 3 (2015) 371– –381

[25] E.M. Monono, J.A. Bahr, S.W. Pryor, D.C. Webster, D.P. Wiesenborn, Org. Process. Res. Dev. 19 (2015) 1683– –1692


368

[26] V.B. Borugadda, V. Goud, J. Energy Eng. 142 (2016) 15010–15020

[27] M. Mushtaq, I.M. Tan, M. Nadeem, C. Devi, S.Y.C. Lee, M. Sagir, U. Rashid, Grases Aceites 64 (2013) 103–114

[28] V.B. Borugadda, V. Goud, J. Cleaner Prod. 112 (2016) 4515–4524

[29] N. Boquillon, C. Fringant, Polymer 41 (2000) 8603–8613 [30] J.M. Pin, N. Sbirrazzuoli, A. Mija, ChemSusChem 8

(2015) 1232–1243

[31] C. Ding, P.S. Shuttleworth, S. Makin, J. Clark, A.S. Matharu, Green Chem. 17 (2015) 4000–4008

[32] M.Y. Shah, S. Ahmad, Prog. Org. Coat. 75 (2012) 248–252

[33] M. Bähr, R. Mülhaupt, Green Chem. 14 (2012) 483–489 [34] Official methods of analysis of the AOAC, 1986 [35] Official Methods and Recommended Practices of the

American Oil Chemists’ Society, American Oil Chemists’ Society, Champaign, IL, 1987

[36] Z.P. Xu, K.T. Chuang, Chem. Eng. Sci. 52 (1997) 3011– –3017

[37] S. Leveneur, J. Wärnå, T. Salmi, D.Yu. Murzin, L. Estel, Chem. Eng. Sci. 64 (2009) 4101–4114

[38] J.V.Jr. de Quadros, R. Guidici, Chem. Eng. Process. 100 (2016) 87–93.

OLGA GOVEDARICA MILOVAN JANKOVIĆ

SNEŽANA SINADINOVIĆ-FIŠER DRAGAN GOVEDARICA

Univerzitet u Novom Sadu, Tehnološki fakultet Novi Sad, Bulevar cara Lazara

1, 21000 Novi Sad, Srbija

NAUČNI RAD

OPTIMIZACIJA EPOKSIDOVANJA LANENOG ULJA PRIMENOM METODOLOGIJE ODZIVNE POVRŠINE

Epoksidovana biljna ulja imaju široku primenu u hemijskoj industriji. Pri proizvodnji ovih ulja potreban je izbor takvih procesnih uslova pri kojima bi se postigao maksimum pri-nosa epoksida. Zbog toga su u ovom radu određene optimalne vrednosti procesnih uslova epoksidovanja lanenog ulja persirćetnom kiselinom formiranom in situ u pri-sustvu jonoizmenjivačke smole kao katalizatora primenom metodologije odzivne povr-šine i Box-Behnken eksperimentalnog plana. Ispitivan je uticaj temperature (65-85 °C), molskog odnosa vodonik peroksida prema 1 molu dvostruke veze lanenog ulja (1,1:1-–1,5:1), količine katalizatora (10-20 mas.%) i reakcinog vremena (5-13 h) na prinos epo-ksida. Analizom varijanse je utvrđeno da je dobijena regresiona jednačina statistički značajna sa koeficijentom determinacije (R2) od 98,95%. Primenom regresione jedna-čine određene su optimalne vrednosti procesnih uslova, i to temperature od 70,6 °C, molskog odnosa vodonik peroksida prema 1 molu dvostruke veze lanenog ulja od 1,5:1, količine katalizatora od 20 mas.% i reakcinog vremena od 7 h. Pri ovim vrednostima procesnih uslova postignut je relativni prinos epoksida od 84,73±0,07%, koji se dobro slaže sa očekivanom vrednošću od 87,6%. Epoksidovano laneno ulje sa visokim sadr-žajem epoksi kiseonika (8,27±0,01%) i niskim jodnim brojem (4,22±0,49) je dobijeno pri-bližno izotermskim šaržnim procesom pri relativno blagim i bezbednim procenim uslovima.

Ključne reči: Box-Behnken eksperimentalni plan; epoksidovanje; laneno ulje; optimizacija; metodologija odzivne površine.

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice

Documents

OLGA GOVEDARICA OPTIMIZATION OF THE ... No3_p357-368...Available on line at Association of the Chemical Engineers of Serbia AChE. Chem. Ind.Chem. Eng. Q. 24 (4) 357−368 (2018) CI&CEQ