Optimization of Performance Model of Ethyl Acetate Saponifi Cation Using Multiple Regression Analysis

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

ISSN 1070-4272, Russian Journal of Applied Chemistry, 2018, Vol. 91, No. 11, pp. 1895−1904.

© Pleiades Publishing, Ltd., 2018.

VARIOUS TECHNOLOGICAL
PROCESSES

Optimization of Performance Model of Ethyl Acetate


Saponification Using Multiple Regression Analysis1
M. K. Al Mesfera, M. Danisha,*, and M. M. Alamb
a Department of Chemical Engineering, King Khalid University, Abha 61411, KSA
b College of Applied Science, King Khalid University, Abha 61411, KSA
*e-mail: danishmhd111@yahoo.com

Received October 2, 2018

Abstract—The purpose of the current study is to optimize the batch reactor performance model using multiple
regression analysis for saponification of ethyl acetate by sodium hydroxide. Reaction temperature, reactor volume,
agitation rate and reactants initial concentration were the main parameters examined including their interaction
effect. Selected process response was the reaction conversion with respect of NaOH. Regression analysis was used
to screen out the insignificant factors and the reaction temperature and volume were found to have insignificant
effect on the response at the 5% selected significance level (α = 0.05). As a result of multiple regression analysis,
agitation rate and reactants concentration were found to be significant operating parameters. The dependence of
reaction conversion (response) on agitation rate and concentration was explained by a second order polynomial
model and it was concluded that regression model with second order polynomial was good enough to fit the ex-
perimental data. The maximum conversion (99.5%) was obtained under optimum operating conditions of agitation
rate (70 rpm) and reactants concentration ( 0.05 M) as evident from surface contours.
Keywords: ethyl acetate saponification, regression analysis
DOI: 10.1134/S1070427218110228

INTRODUCTION The computational efforts, number of experimental


runs needed to analyze the behavior of model or its
Multiple regression analysis is a powerful technique validation; time factor and cost can be minimized by
used to predict the unknown value of variable from the developing a systematic procedure. A study devoted for
value of two or more variables- also called predicators. process improvement to minimize the process time has
The variable whose value is to be predicted is called as been conducted by application of factorial design [3].
dependent variables and ones whose values are used Statistical tools have been applied for organized and
for prediction are known as independent variables. In systematic procedure to a batch reactor [4]. It has been
scientific research, experimental work is essential in the reported that advancements in the design of experiments
development of process, selection and optimization. The for last many years has made it possible to apply the
effect of factors on the response is explained by regression technique for complex applications chemical/biochemical
analysis. To improve the selection, development and kinetics, heat/mass transfer operations and regression
optimization of process, statistical designs are extensively modeling. Recently the studies reported on enhancement
used. Statistical design of experiment is a well known of naphtha steam cracking [5], process improvement of
technique used in various fields such as chemical and ethyl acetate saponification [6] and recipe improvement
biological fields, food industry, etc., to produce high of batch reactor [7], all focused on the confidence level
quality desired products economically, to ensure more attained with design of experiments.
sound and reliable process [1, 2]. The selection of method of statistical design
1 The text was submitted by the authors in English. of experiments firmly depends on the purpose of

1895
1896 AL MESFER et al.

experimentation [2, 8]. The simple and most widely After screening experiment, the agitation rate and initial
designs optimizations are screen design and response reactants were found as significant factors and optimum
surface methodology (RSM). The objective of screening values of significant factors were determined using second
design is to screen out the insignificant factors and then order polynomial by obtaining surface contours.
RSM is applied to optimization of significant factors.
In this study, saponification of ethyl acetate by sodium EXPERIMENTAL
hydroxide using batch reactor has been examined for Reaction Kinetics
process improvement. The multiple regression technique
The saponification reaction of ethyl acetate can be
was applied to optimize the polynomial regression model.
represented in terms of stochiometric coefficients [6].
The hydrolysis of ethyl acetate (CH3COOC2H5) by
sodium hydroxide (NaOH) to produce sodium acetate
NaOH + CH3COOC2H5 → CH3COONa + C2H5OH, (1)
(CH3COONa) and ethyl alcohol (C2H5OH) is known as
saponification though the end product is not soap. Sodium This irreversible reaction is a second –order overall
acetate obtained in this study is commercial valuable but first order with respect to each reactant and the rate
carboxylic salt, is used widely in industries where ethanol expression is given by Eq. (2) [6].
as by-products, can be used as biofuel.
The most of the previous studies [9–15] in literature –rNaOH= ̶ rEtOAc= kCNaOHCEtAc. (2)
have concentrated on kinetics and reaction mechanics
of ethyl acetate saponification. It was suggested that During this reaction, hydroxyl ions are consumed and
saponification of ethyl acetate proceeds through direct acetate ions are produced. A decrease in conductivity is
attack of hydroxyl ion carbon atom of ethyl acetate noticed as the reaction progresses since hydroxide ions are
[16]. A very mild and fast saponification method was more conductive than the acetate ions. The conductivity
established which utilizes dichloromethane/methanol change is used to monitor the progress of saponification.
(9 : 1) as solvent with low NaOH concentration at room A relationship between the conductivity of the reaction
temperatures [17]. mixture and sodium hydroxide concentration is obtained
A modeling and simulation study of ethyl acetate as presented by Eq. (3) [6]. Investigators [9–11, 16] have
saponification has been conducted [18]. Studies devoted specifically focused on kinetics and reaction mechanics
to the optimization of ethyl acetate using two-level of saponification of ethyl acetate by sodium hydroxide.
factorial design have been reported [19, 20]. A potential C – C∞ СNAOH – СNAOH∞
study devoted on process intensification using fixed –––––– = –––––––––––––––, (3)
C0 – C∞ СNAOH,0 – СNAOH∞
point method of design with objectives of reducing
the production of undesired byproducts, improving the Since, CNaOH,∞ → 0 as t → ∞ Eq. (3) can be rewritten
reaction conversion and yield and improving the energy as given in Eq. (4)
efficiency has been reported [21].
СNAOH C – C∞
In this study the hydrolysis of ethyl acetate has been –––––– = 1 – XNaOH = –––––––, (4)
investigated using multiple regression analysis. This СNAOH,0 C0 – C∞
study specifically focused on the process improvement
of ethyl acetate saponification. This study consists of two EXPERIMENTAL
stages: screening experiments aimed to screen out the
insignificant factors and RSM. The reaction conversion The experimental data obtained [22] during the ethyl
of NaOH (Xa) was selected as response for analyzing the acetate saponification has been used to carry out the
system performance. All the possible factors thought to current optimization study using multiple regression.
have effect on response and it was decided to examine AR grade sodium hydroxide (NaOH) and ethyl acetate
the main and interaction effect of agitation rate (rpm), (CH3COOC2H5) were used to conduct the experiments.
reaction temperature (K), rector volume (L) and reactants The stock solutions of desired concentration of sodium
concentrations (M).The experimental results and findings hydroxide and ethyl acetate were prepared using high
[22] were used to carry out the current optimization study. purity distilled water. Both sodium hydroxide and sodium

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018


OPTIMIZATION OF PERFORMANCE MODEL OF ETHYL ACETATE SAPONIFICATION 1897

Table 1. Range of operating variables (predictors) reactor volume and reactants concentration. In first stage,
Factors Max Center Min the insignificant factors were screen out. The range of
operating conditions selected for design of experiments
Agitation rate X1, rpm 150 110 70
is shown in Table 1 [22]. In second stage, the significant
Temperature X2, °C 40 30 20 factors were used to optimize the reaction conditions
Reactor volume X3, L 1.8 1.5 1.2 that maximize the conversion. The optimum factors
Reactants were regressed into second order polynomial model. The
0.075 0.05 0.025
concentration X4, M maximum and minimum ranges of operating variables
used for analysis are shown in Table 1.

acetate contribute conductance to the hydrolysis reaction RESULTS AND DISCUSSION


where ethyl acetate and ethanol do not. Experimental Effects
The conductivity of sodium hydroxide solution at a The following findings of the experimental
fixed concentration and temperature is not same as that investigation [22] were used to carry out the design of
of sodium acetate at the same molarity and temperature. experiments. These may be summarized as:
During a chemical reaction, the conductivity of reaction
(1) The conversion of NaOH increases with increased
solution changes as more of the reactants are converted. A
reaction temperature under the studied range of reaction
study devoted to the conductivity measurement technique
temperatures until the stable condition is achieved.
for saponification of ethyl acetate has been reported [23].
(2) Reaction conversion decrease with increased
Plan of Experiments volume of reaction mixture within the studied range.
(3) The conversion of sodium hydroxide declines
The experiments were conducted in two stages, with increased agitation rate from 70 rpm to 150 rpm.
screening and optimization. Four factors were considered (4) Higher initial reactant concentration leads to
for analysis i.e., agitation rate, reaction temperature, decreased reaction conversion.

Table 2. Correlations analysisa

Model X1 X2 X3 X4 XA
X1 Pearson Correlation 1 0.000 0.048 –0.019 –0.483
Sig. (2-tailed) 1.000 0.859 0.943 0.058
N 16 16 16 16 16
X2 Pearson Correlation 0.000 1 0.000 0.000 0.139
Sig. (2-tailed) 1.000 1.000 1.000 0.607
N 16 16 16 16 16
X3 Pearson Correlation 0.048 0.000 1 0.017 –0.331
Sig. (2-tailed) 0.859 1.000 0.949 0.211
N 16 16 16 16 16
X4 Pearson Correlation –0.019 0.000 0.017 1 –0.279
Sig. (2-tailed) 0.943 1.000 0.949 0.295
N 16 16 16 16 16
XA Pearson Correlation –0.483 0.139 –0.331 –0.279 1
Sig. (2-tailed) 0.058 0.607 0.211 0.295
N 16 16 16 16 16
a (X ) Agitation rate, rpm; (X2) Reaction temperature, °C; (X3) Reactor volume, L; (X4) Reactants concentration, M, (XA) Reaction conversion.
1

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018


1898 AL MESFER et al.

Table 3. Model summary with four predictorsa


Std. Error of the Change statistics
Model R R square Adjusted R square
estimate R square change F change df1 df2 Sig. F change
1 0.854 0.827 0.819 0.1464736 0.427 2.053 4 11 0.156
a Predictors: (Constant), X1, X2, X3, X4.

Table 4. ANOVA with four predictorsa


Model Sum of squares df Mean square F Sig.
1 Regression 0.176 4 0.044 2.053 0.156
Residual 0.236 11 0.021
Total 0.412 15
a Dependent variable: X . Predictors: X X , X X .
A 1, 2 3, 4

Table 5. Coefficients with four predictorsa


Unstandardized coefficients Standardized coefficients
Model t Sig.
B std. error beta
1 Constant 1.945 0.612 3.176 0.009
X1 –0.005 0.002 –0.474 –2.074 0.062
X2 0.006 0.009 0.139 0.610 0.554
X3 –0.406 0.306 –0.303 –1.327 0.212
X4 –4.863 3.921 –0.283 –1.240 0.241
a Dependent variable: XA.

Table 6. Residuals Statistics with four predictorsa

Minimum Maximum Mean Std. Deviation N


Predicted value 0.528988 0.973730 0.702575 0.1083841 16
Residual –0.2215461 0.1559357 0E–7 0.1254325 16
Std. Predicted value –1.602 2.502 0.000 1.000 16
Std. Residual –1.513 1.065 0.000 0.856 16
a Dependent variable: XA.

Screening Experiments slightly. There is inverse relationship between reactants


Correlation analysis was used to check the association concentration and reaction conversion as depicted in
among the operating variable and reaction conversion Table 2 which states that percentage increase in reactants
as depicted in Table 2. There is negative moderate rela- concentration resulted to 27.9% decline in reaction
tionship between agitation rate and reaction conversion. conversion and vice versa
The coefficient of correlation between agitation rate and Step wise multiple regression analysis was used to
reaction conversion was –0.483 which indicates that one optimize the reaction conversion by using the operating
percent changes in agitation rate resulted to 48.3% change variables as mentioned in Table 1. The first steps of
in reaction conversion. Whereas reaction temperature and regression analysis were depicted in Tables 3–6. To
reaction conversion have low direct correlation which optimize the conversion reaction by using these four
states that increase in reaction temperature resulted to operating variables, R value in Table 3 demonstrates that
slightly increase in reaction conversion. Negative cor- there is strong high positive association between operating
relation was found between reactor volume and reaction variables and reaction conversion. R square was very high
conversion. which states that model is explaining 82.7% fitness to
The relationship states that percentage change in optimize the reaction conversion. Adjusted R square states
reactor volume resulted decrease in reactor conversion that the contribution of extraneous variables to maximize
RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018
OPTIMIZATION OF PERFORMANCE MODEL OF ETHYL ACETATE SAPONIFICATION 1899

Table 7. Model summary with two predictorsa

Std. error of the Change statistics


Model R R square Adjusted R square
estimate R square change F change df1 df2 Sig. F change
1 0.863 0.717 0.611 0.1472149 0.317 3.010 2 13 .084
a Dependent variable: XA; Predictors: (Constant) X1, X4

Table 8. ANOVA with two predictorsa

Model Sum of squares df Mean square F Sig.

1 Regression 0.130 2 0.065 3.010 0.084


Residual 0.282 13 0.022
Total 0.412 15
a Dependent variable: XA; Predictors: (Constant) X1, X4

the reaction conversion was 18% approximately. The Y = 1.945 – 0.005X1 + 0.006X2 – 0.406X3
significance F change which is termed as p value of – 4.863X4 + e. (6)
the regression model, was 0.156 demonstrating that the
contribution of all four predictors differs significantly The residual analysis as depicted in Table 6 represents
with 95% level of significance. the maximum and minimum reaction conversion by
Following regression equation was developed to predicted by this model [Eq. (6)]. As reported in the
maximize the reaction conversion. table maximum reaction conversion is 0.9737, minimum
reaction conversion is 0.5289 while average reaction
Y = β0 + β1X1 + β2X2 + β3X3 + β4X4 + e, (5) conversion is 0.7026 which indicates that model is still
not providing 100% results and it can further be improved
where Y is a dependent variable as reaction conversion, with addition and deletion of some predictors.
X1, X2, X3, and X4 are independent variables as agitation To improve the reaction conversion (output), reaction
rate, reaction temperature, reactor volume and reactants temperature X2 and reactor volume X3 were removed
concentration respectively. β0 is intercept of the model from the existing model and considered as insignificant
and considered as constant parameter while β1, β2, β3, and
factors on the basis of coefficient and their significant
β4 are coefficient of independent variables and e is used
values. The regression analysis has been repeated with
as error term in the model.
the two predictors i.e. agitation rate (X1) and reactant
As reported in ANOVA Table 4, p value is greater concentration (X4) as depicted in Table 7.
than 0.05 illustrating that the predictors’ variances of
contribution are significant statistically. Calculated Optimization of conversion using significant fac-
F value is more than the tabulated value and it also tors. To optimize the regression model, analysis has been
signifies that the predictors’ variances of contribution are repeated by considering only two significant factors,
significant statistically. i.e., agitation rate (X1) and reactants concentration (X4).
The contribution of all predictors on process response Optimization of conversion was performed using only
(reaction conversion) is depicted in Table 5. Sig. p value is significant factors and model summary is depicted in
very high of all predictors except constant which explain Table 7. As observed from the table from the value of R
that the intercept of predictor are highly significant at is 0.863, predictors have strong positive correlation with
95 per cent significant level. Following Eq. (6) has been the reaction conversion as indicated by the value of R.
developed with the help of coefficients (Table 5). As R square (0.717) showed the fitness of model decline from
reported in following equation contribution of reaction 0.827 with four predictors to 0.717 with two predictors.
temperature and reactor volume is minimal among all Adjusted R square is 0.611 which demonstrates that the
variables in predicting the reaction conversion (output). contribution of extraneous variable increased from 18

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018


1900 AL MESFER et al.

Table 9. Coefficients with two predictorsa


Unstandardized coefficients Standardized coefficients
Model t Sig.
B std. error beta
1 Constant 1.560 0.352 4.430 0.001
X1 –0.005 0.002 –0.489 –2.130 0.053
X4 –4.958 3.940 –0.289 –1.258 0.230
a Dependent variable: X
A

Table 10. Residuals Statistics with two predictorsa


Parameter Minimum Maximum Mean Std. Deviation N
Predicted Value 0.556540 0.98772 0.702575 0.0932621 16
Residual –0.2119748 0.2073033 0E–7 0.1370496 16
Std. predicted value –1.566 2.886 0.000 1.000 16
Std. residual –1.440 1.408 0.000 0.931 16
a Dependent variable: X
A

Table 11. Model Summary of 2nd order polynomial with two predictorsa

Std. Error of the Change statistics


Model R R square Adjusted R square
estimate R square change F change df1 df2 Sig. F change
1 0.963 0.889 0.866 0.1513561 0.389 1.748 4 11 0.209
a Dependent variable: XA Predictors: (Constant) X1, X4, X12, X42

Table 12. ANOVA of 2nd order polynomial with two predictorsa

Model Sum of squares df Mean square F Sig.


1 Regression 0.160 4 0.040 1.748 0.209
Residual 0.252 11 0.023
Total 0.412 15
a Dependent variable: XA; Predictors: (Constant), X1, X4, X12, X42.

per cent with four predictors to 39% with two predictors. There was an increase in minimum and maximum
The significance F change which is termed as p value of conversion with two predictors as compared to reaction
the regression model was 0.084 demonstrating that the conversion with four predictors. Regression model with
contribution of all four predictors differs significantly two predictors (significant variables) provides higher
with 95 per cent level of significance. value of conversion as compared to the regression model
As reported in ANOVA Table 8, sig. p value is greater with four predictors.
than 0.05 which shows that the variances of predictors to
Y = 1.560 – 0.005X1 – 4.958X4 + e (7)
the reaction conversion differs significantly with 5% level
of significance and F values also validate that the variances
As observed from Table 10 on screening out the two
of predictors to the reaction conversion varies significantly. least contributor predictors, correlation between predictors
From the coefficient of Table 9, equation (7) was and reaction conversion has improved but fitness of model
extracted as regression equation with two predictors for decline and maximum reaction conversion has increased.
reaction conversion. The following equation provides It can be concluded that this model is giving improved
maximum reaction conversion as 0.9877, minimum conversion but model is not significant which indicates
reaction conversion as 0.5565 while average reaction that the model can be improved further to optimize the
conversion of 0.7026 as depicted in residual Table 10. reaction conversion.
RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018
OPTIMIZATION OF PERFORMANCE MODEL OF ETHYL ACETATE SAPONIFICATION 1901

Table 13. Coefficients of 2nd order polynomial with two predictorsa


Unstandardized coefficients Standardized coefficients
Model t Sig.
B Std. Error Beta
1 (Constant) 1.398 1.218 1.148 0.275
X1 –0.011 0.020 –1.134 –0.551 0.592
X4 14.383 18.864 0.837 0.762 0.462
X12 0.000031 0.000 0.674 0.328 0.749
X42 –194.790 185.206 –1.154 –1.052 0.315
a Dependent variable: X
A

Table 14. Residuals Statistics 2nd order polynomial with 2 predictorsa


Parameter Minimum Maximum Mean Std. Deviation N
Predicted value 0.446273 0.995018 0.702575 0.1033476 16
Residual –0.2055206 0.1923794 0E–7 0.1296136 16
Std. predicted value –2.480 2.830 0.000 1.000 16
Std. residual –1.358 1.271 0.000 0.856 16
a Dependent variable: XA

Optimization of model using second order polyno- coefficient beta values to get the optimum solution of
mial. To optimize the reaction conversion it was decided ethyl acetate saponification. It can be concluded from
to construct second order polynomial model between the equation 9 that 100 percent increase in agitation rate
significant factors to investigate the effect of significant will reduce the conversion by 1.1%. Concentration shows
factors (agitation rate, X1 and reactants concentration, X4) positive relationship with the conversion as depicted in
on response (Conversion, Xa) as follows the Table 13. The second order condition of agitation
rate is positively related with the response while second
Y = βo + β1X1 + β4 X4+ β16 X12+ β19X42 + e, (8) order condition of concentration has negative impact on
dependent variable.
β16 and β19 are the coefficient of independent variable
while X12 and X42 are the square terms of their respective Y = 1.398 – 0.011X1 + 14.383X4+ 0.000031X12
predictors. To find out the optimum conversion using – 194.709X42 + e. (9)
second order polynomial, regression analysis was used
as depicted in Table 11. As reported in the table there As depicted in residual Table 14, the maximum
is strong positive relationship between the predictors, conversion was 99.5% while the minimum was 44.6%
their polynomial and conversion. The observed R square with an average conversion rate 70.3%. The 99.5%
explained that model is fitting around 90% while adjusted reaction conversion predicted by Eq. (9) is the highest
R square explain the contribution of external factors is among all three models. The variation in conversion
around 13% which is very less as compared to the previous is around 10% in this model as depicted. The highest
findings. The sig. F change is 0.209 greater than 0.05 and variation was found in the first model with four predictors
it is not significant with 5% level of significance. while the lowest variation was observed in second model
with two predictors. Moreover, in this polynomial model
Second order polynomial with two predictors explain
it is slightly increased but this variation is very minor
that the F value is very less and the significance level is and provides maximum conversion. It can be concluded
more than 0.05 which indicates that the contribution of that the reaction conversion is maximum with the use of
each predictors do not vary significantly with 5% level significant factors (i.e. agitation rate and concentration)
of significance as depicted in Table 12. and their second order conditions. The combination of
The coefficient Table 13 was used to develop predictors with conversion which provides maximum
the following polynomial regression equation using conversion is represented in Fig. 1. As evident from the

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018


1902 AL MESFER et al.

Estimated marginal means of XA Conversion


1.00 1.00
X4 Observed
0.03 Quadratic
0.90 0.04
0.05 0.90
0.08
0.80
0.80

0.70
0.70
0.60
0.60
0.50
0.50
0.40
90.00 110.00 130.00 150.00 0.40
X1 90.00 80.00 100.00 120.00 140.00 160.00
Non-estimable means are not plotted Agitation rate
Fig. 1. Optimization of conversion as a function of significant Fig. 2 Optimization of conversion using agitation rate.
variables.

0.075
Ymax
Reactant concentration, M

x = 126.24
y = 0.05 Ymin
0.025
70 Xmin Xmax 150
Fig. 3. Response Surface Contours.

Fig. 1, the maximum conversion is 99.5 percent at an obtained at high agitation rate of 150 rpm. It validates the
optimum operating conditions of agitation rate (70 rpm) experimental findings of study devoted to saponification
and reactants concentration (0.05 M). of ethyl acetate by sodium hydroxide.
Figure 2 represents the optimization of fractional To validate the finding of regression Eq. 9 for
reaction conversion using different combination of optimization of conversion using agitation rate and
an agitation rate as depicted. It is observed that the concentration, response surface contours were obtained
conversion is maximum i.e. 99.5 percent at an agitation by using the same regression equation at different levels
rate of 70 rpm and decreases with increased rate of an of reactants concentration and agitation rate as shown
agitation. The minimum conversion of 44.6 percent was in Fig. 3. The maximum predicted conversion of 99.5%
RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018
OPTIMIZATION OF PERFORMANCE MODEL OF ETHYL ACETATE SAPONIFICATION 1903

Normal P‒P plot of regression studentized residual


Dependent variable: Conversion
1.00 1.5

1.0

Regression studentized residual


0.80
0.5
Expected Cum Prob

0.60
0.0

0.40 ‒0.5

0.20 ‒1.0

‒1.5
0.00
0.0 0.2 0.4 0.6 0.8 1.0 ‒1.5 ‒1.0 ‒0.5 0.0 0.5 1.0 1.5
Observed Cum Prob Regression studentized residual
Fig. 5. Scatter plot of conversion (output) at different
Fig. 4. Regression standard residual for optimization. experiments.

was obtained at optimum conditions of agitation rate analysis with four predictors was used to identify the
(70 rpm) and reactants concentration (0.05 M). insignificant factors and the maximum conversion of
Figure 4 is the plot of regression standardized 97.4 percent was observed.
residual for optimization of conversion mechanism At the second stage linear regression was used
during experiment. Equal variance was observed during with only two significant predictors (agitation rate and
all the experiment as depicted in the same figure. It can concentration) to obtain the maximum conversion. The
be concluded that there will be no significant difference maximum conversion of 98.8 % was found with two
between expected and observed optimization of significant predictors. In order to further optimize the
conversion process. conversion, second order polynomial was used with
Figure 5 represents scatter plot of conversion and two significant predictors. The initial concentration of
demonstrate that the experimental conversion are on reactants and agitation rates and their quadratic effects
the straight line except few values which shows that were found to be significant factors. Maximum conversion
the random error are minimum and constant variance. as 99.5 percent was observed at an optimum condition
The graph also highlights the number of outlier during of agitation rate (70 rpm) and reactant concentration
experiments for optimization of reaction conversion. (0.05 M) and their polynomials shown in contour plots.

REFERENCES
CONCLUSIONS
1. Montgomery, D.C., Design and Analysis of Experiments,
The optimization study to improve the performance
New York: John Wiley and Sons, 1991, pp.270–569.
model of ethyl acetate saponification has been carried out
with respect to consumed material and obtained yield. 2. Zivorad, R.L., Design of Experiments in Chemical
Engineering, Weinheim : Wiley-VCH, 2004.
In this study, the work was divided into two sections-
screening out the insignificant factors and optimizing 3. Carr, J.M. and McCraken, E.A., Chem. Eng. Prog., 1960,
the significant factors. As a result of the screening vol.56 (11), pp. 56–61.
experiments, reaction temperature (range 20–40°C) 4. Asprey, S.P. and Macchietto, S., Comput. Chem. Eng.,
and reactor volume (range: 1.2–1.8 L) were found to 2000, vol. 24, pp. 1261–1267.
be insignificant operating parameters. Linear regression 5. Keyvanloo, K., Towghi, J., Sadrameli, S.M., and

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018


1904 AL MESFER et al.

Mohamadalizadeh, A., J. Anal. Appl. Pyrolysis, 2010, pp. 46–50.


vol. 87, p. 224–230. 15. Schneider, M.A. and Stoessel, F., Chem. Eng. J., 2005,
6. Bursali, N., Ertunc, S., and Akay, B., Chem. Eng. Process., vol. 115, pp.73–83.
2006, vol. 45, pp. 980–989. 16. Tsujikawa, H. and Inoue, I., Bull. Chem. Soc. Jpn, 1966,
7. Zofia, V.L., Comput. Ind., 1998, vol.36, pp. 279–300. vol. 39(9), pp.1837–1842.
8. Atkinson, A.C., Bagacka, B., and Bogaki, M.B., Chemom. 17. Theodorou, V., Skobridis, K., Tzakos, A.G., and
Intell. Lab Syst., 1998, vol. 43, pp.185–198. Ragoussis, V., Tetrahedron Lett., 2007, vol. 48, pp. 8230–
9. Mata-Segreda, J.F., J. Am. Chem. Soc., 2002, vol. 124 8233.
(10), pp. 2259–2262. 18. Wijayarathne, U.P.L. and Wasalathilake, K.C., J. Chem.
10. Ortiz, M.I., Romero, A., and Irabien, A., Thermochimica Eng. Process Technol., 2014, vol. 5(6), pp. 1–8.
Acta, 1989, vol. 141, pp.169–180. 19. Ahmad, A., Ahmad, M.I., Younas, M., Khan, H., and
11. Garu, M.G., Nougues, J.M., and Puigjaner, L., 2002. Chem. Shah, M.H., Iran. J. Chem. Chem. Eng., 2013, vol. 32(4),
Eng. J. 88, pp. 225–232. pp. 33–44.
12. Danish, M., Al Mesfer, M.K., and Rashid, M.M., Int. J. 20. Ullah, I., Ahmad, M.I., and Younas, M., Pak. J.Eng. &
Eng. Res., 2015, vol. 5(2), pp.74–78. Appl. Sci., 2015, vol. 16, pp. 84–92.
13. Kuheli, D., Sahoo, P., Saibaba, M., Murali, N., and 21. Li, H., M, Ying., Li, Xingang., and Gao, X., Chem. Eng.
Swaminthan, P., Int. J. Chem. Kinet., 2011, vol. 43, Res. Des., 2016, vol.111, pp. 479–491.
pp.648–656. 22. Al-Mesfer, M., Int J. Chem. Reactor Eng., 2017.
14. Mukhtar, A., Shafiq, U., Khan, A.F., Qadir, H.A., and doi: 10.1515/ijcre-0174
Qizilbash, M., Res. J. Chem. Sci., 2015, vol. 5(11), 23. Walker, J., Proc. R Soc A, 1906, pp. 78, pp. 157–160.

RUSSIAN JOURNAL OF APPLIED CHEMISTRY Vol. 91 No. 11 2018

You might also like