You are on page 1of 21

Food Bioprocess Technol (2017) 10:413–433

DOI 10.1007/s11947-016-1855-2

REVIEW

Applications of Response Surface Methodology in the Food


Industry Processes
Mahmoud Yolmeh 1 & Seid Mahdi Jafari 1

Received: 8 August 2016 / Accepted: 19 December 2016 / Published online: 11 January 2017
# Springer Science+Business Media New York 2017

Abstract Response surface methodology (RSM) is a tech- increasing cost and time. Finding a condition with the best
nique widely used to optimize various processes. This review output for a system is the main purpose of Boptimization^
presents the state-of-the-art applications of RSM in the opti- which has been commonly used (Baş & Boyacı, 2007). In
mization of different food processes such as extraction, dry- the past, optimization in food processes has been performed
ing, blanching, enzymatic hydrolysis and clarification, pro- through investigating the effect of one-parameter changes
duction of microbial metabolites, and formulation. The prin- on a response while all others are held at a constant level
ciples of RSM, its implementation steps, and different designs (Bezerra et al., 2008). The main disadvantages of this meth-
(full factorial design (FFD), Box-Behnken design (BBD), and od are the interactive effects among the variables, which are
central composite design (CCD)) are described. Furthermore, not considered and there is a lack of explanation of the
this work presents a comprehensive study of RSM literature complete effect of the factors on the response. In addition,
recently published about the various food process fields and this method increases the number of experiments needed to
evaluating their RSM elements summarized in tables. Finally, conduct the research, which leads to increased cost and time
the challenges and future prospects of using this statistical (Baş & Boyacı, 2007; Bezerra et al., 2008).
technique in the food industry processes are discussed. It can In order to resolve this problem, optimization studies can
be concluded that appropriate selection of RSM design, inde- be carried out by using multivariate statistic methods.
pendent variables (screening), and levels of the factors signif- Response surface methodology (RSM) is the most popular
icantly influences the successful application of RSM. In addi- multivariate statistic technique, which has been used in the
tion, validity evaluation of the optimum conditions predicted optimization of food processes (Baş & Boyacı, 2007). RSM
through RSM is crucial too. is a collection of statistical and mathematical methods
established on the fit of a polynomial model to the data that
must depict the behavior of a data set with the purpose of
Keywords Response surface methodology . Extraction .
making statistical predictions. The approach is useful for op-
Drying . Biotechnology . Optimization . Food industry
timizing, designing, developing, and improving processes
where a response or responses are affected by several
variables (Ghorbannezhad et al., 2016; Kaushik et al., 2006).
Introduction Before using the RSM approach, the correct experimental
design must be selected that will designate which treatments
The food industry is seeking to improve system perfor- should be done in the experimental region being studied. For
mance and to enhance process efficiency without this purpose, experimental designs for quadratic response sur-
faces, such as three-level factorial, central composite, and
* Seid Mahdi Jafari
Box-Behnken, should be applied (Bruns et al., 2006).
smjafari@gau.ac.ir Generally speaking, RSM is an appropriate approach and
widely used to optimize food industry processes. However,
1
Faculty of Food Science and Technology, University of Agricultural some researchers are not familiar with this approach. This
Sciences and Natural Resources, Gorgan, Iran misusing can be attributed to perform inappropriate
414 Food Bioprocess Technol (2017) 10:413–433

experimental design, inappropriate screening of independent step of optimizing. This step can be done using full or frac-
variables, and improper selection of levels for independent tional factorial designs (for 2–4 factors) and fractional facto-
variables. We have tried to address these issues in this study. rial or Plackett-Burman design (for 5 or more factors). In such
In addition, this paper presents a comprehensive summary of designs, only main effects are estimated; interactions between
literature recently published on optimization of food processes independent variables are usually considered insignificant and
by using RSM. BCan RSM be utilized for optimization of all are neglected. The Plackett-Burman design type is a two level
food industry processes without any limitation?^ Before ad- fractional factorial screening design for studying N-1 vari-
dressing this challenge, the principles of the approach are first ables using N runs, where N is a multiple of 4. The screened
presented. factors then are provided to RSM for optimization (Lundstedt
et al., 1998; Myers et al., 2016).

Theory and Steps of RSM


Selection of a Regression Model
The RSM approach was developed by Box and colleagues in
the 1950s, with the major advantage being the reduced num- The least square multiple regression methodology is usually
ber of experiments that are required to appraise multiple pa- applied to investigate the relationship between the indepen-
rameters and their interactions (Bruns et al., 2006; Ge et al., dent and dependent variables. A multiple regression equation
2002). The term BResponse Surface Methodology^ was de- can be used to fit the second-order polynomial equation based
rived from the graphical view created after fitting the mathe- on the experimental data as follows:
matical model. RSM depicts the effect of the independent
variables on the dependent variables. In addition, the approach y ¼ β0 þ ∑ki¼1 βi X i þ ∑ki¼1 βii X 2i þ ∑ki≠ j¼1 βij X i X j þ ε
also generates an empirical model.
where y represents the predicted response, β0 is the model
Before beginning the discussion on the principles and ap-
intercept, and βi, βii, and βij are the regression coefficients
plications of RSM in optimization of food processes, it is
for the linear, quadratic, and interactive effects of the mod-
necessary to introduce and clarify some important terms.
el, respectively. Xi and Xj are the factors and k is the num-
BExperimental domain^ is the experimental field that will
ber of factors (Ghorbannezhad et al., 2016). The model
be investigated, which is defined by the maximum and mini-
used for optimizing food processes in the RSM meth-
mum limits of the independent variables. BExperimental
odology is not always a second order. The regression
design^ is a specific system of experiments defined by a ma-
coefficients that have a statistically significant effect
trix created with the different level combinations of the inde-
on response should be considered in the equation.
pendent variables. Central composite, Box-Behnken, and
Therefore, if quadratic terms are not statistically signif-
Doehlert designs are examples of experimental designs.
icant, those coefficients are removed and equation will
BIndependent variables^ or Bfactors^ are input variables
be linear (Myers et al., 2016).
which can be changed independently of each other. BRuns^
are series of tests that form an experiment. BDependent
variables^ or Bresponses^ are output variables which are in-
Coding the Factor Levels
fluenced by several independent variables. BResidual^ is the
difference between the experimental and calculated result for a
After screening for the important factors and selection of the
determinate set of conditions. A low value of this index is
regression model, the levels of factors are defined. This step is
necessary for a good mathematical model fitted on experimen-
very important and the success of optimizing the process de-
tal data (Bezerra et al., 2008).
pends on this step. Given that, the factors have different units
In the following section, a summary of RSM principles has
and ranges; before executing a regression analysis, the factors
been presented. For more information, readers are referred to
must be normalized. Codification of the levels of factors re-
Bezerra et al. (2008) and Myers et al. (2016).
quires converting each studied value into a range without di-
mensions (−1 to +1). A studied value (zi) can be converted to a
Screening of Independent Variables
coded value (xi) by the following formula:
 
A wide range of factors usually affect the food industry pro- xi ¼ zi −zi 0 =Δzi βd
cesses; therefore, the evaluation of individual effect of each
factor is impossible due to high cost and number of runs need- where Δzi represents the difference from the real value in the
ed. To avoid this, the researcher might choose to first use a higher or lower value from the real value at the central point
Bscreening design^ to recognize those factors which have (zi0), and βd represents the major coded limit value of each
large effects on the response. Indeed, the screening is the first variable (Bezerra et al., 2008).
Food Bioprocess Technol (2017) 10:413–433 415

Verification of the Fitted Model Possibilities are including missing variables and/or higher-
order term of a variable in the model to explain the curvature,
It is possible to calculate the estimated response using the and/or interaction between terms already in the model.
model equation and regression coefficients. However, verifi- Identifying and fixing the problem so that the predictors
cation of the model adequacy must be checked to determine if now explain the information that they missed before should
it is appropriate or not. Several techniques are available, such produce a good-looking set of residuals (Myers et al., 2016).
as prediction error sum of squares (PRESS) residuals, the Various criteria (R2, AAD, residual analysis, and lack-
lack-of-fit test, and residual analysis. The coefficient of deter- of-fit) are used to evaluate the suitability of response sur-
mination (R2) is used to evaluate the general predictive capa- face models for fitting the data, which will be reviewed
bility of the fitted model. However, R2 index alone cannot here. Nearly in all RSM studies, R2 has been used to eval-
demonstrate the accuracy of the model, because the index is uate the general predictive capability of the fitted model.
a measure for amount of the decreasing changeability of re- However, R2 is not a sufficient index for this evaluation
sponse achieved by using the repressor variables in the model. and other criteria should be investigated. Residual analysis
In addition, increasing a variable to the model will increase R2 and lack-of-fit evaluations have not been applied in some
without considering the statistical significance of the addition- papers (as shown in Tables 2, 3, and 4); however, lack-of-
al variable. Thus, a high value of R2 is not necessarily indic- fit value was inappropriate in some of them (the P value of
ative of the accuracy of the model. The absolute average de- lack-of-fit should be insignificant (P > 0.05) for a suitable
viation (AAD) is a better measure of accuracy. AAD is the fitting of response surface models). Unfortunately, AAD
average of the absolute deviations from a central point and is a criterion has not been considered in any of the papers
summary statistic of statistical dispersion or variability, which reviewed here even though it is a useful value to assay
is calculated by the following formula: the suitability of fitting the response surface model, as
mentioned earlier (Ghorbannezhad et al., 2016).
 p .
AAD ¼ ∑ jyi exp–yi calj=yi exp p  100 Graphical Presentation of the Model
i¼1

where p, yi exp, and yi cal represent the number of experi- The visualization of the predicted model equation can be ap-
ment, experimental, and calculated responses, respectively. peared by the response contour and surface plots, as shown in
Both R2 and AAD should be evaluated to check the accu- Fig. 2. These plots are theoretical three-dimensional outputs of
racy of the model. R2 must be near 1 and the AAD between the RSM approach, which indicate the relationship between
the estimated and observed data must be as low as possible. the dependent and independent variables. The response con-
The suitable values of R2 and AAD imply that the fitted tour and surface plots are used to describe changes in response
model depicts the correct behavior of the process and it by changes in the independent variables. Contour plot is a
can be successfully used for optimization of food processes two-dimensional screen of the surface plot, in which, ranges
(Ghorbannezhad et al., 2016). of constant dependent variables is drawn in the plane of the
Moreover, residual analyses (plots of observed-predicted independent variables. Indeed, the contour plots improve un-
values versus the factors and responses) would be a valuable derstanding the shape of a responses surface. As the target
criterion for this assessment. Using residual plots, you can point (minimum, maximum, or in ranged point) is located in
assess whether the observed error (residuals) is consistent with the center of the system (experimental region), the contour
stochastic error. The residuals should not be either systemati- plots show a circle or ellipses status. When the target point
cally high or low. In addition, the residuals should be centered is neither a maximum nor a minimum point, the contour plots
on zero throughout the range of fitted values, as is obvious in show parabolic or hyperbolic system contours. It must be
Fig. 1a. In other words, the model is correct on average for all noted that these plots display visualization of estimated
fitted values. Further, in ordinary least squares context, ran- responses, not the real responses (Baş & Boyacı, 2007;
dom errors are assumed to produce residuals that are normally Myers et al., 2016).
distributed. Therefore, the residuals should fall in a symmet-
rical pattern and have a constant spread throughout the range Determination of Optimal Conditions
(Myers et al., 2016).
Figure 1b shows a problematic residual plot that is unde- It is possible to calculate the optimum point (maximum or
sirable. The non-random pattern in the residuals indicates that minimum point) through the first derivate of the mathematical
the deterministic portion (predictor variables) of the model is equation which depicts the response surface and equates it to
not capturing some explanatory information, which is zero:
Bleaking^ into the residuals. The graph could represent several
ways in which the model is not explaining all that is possible. y ¼ β 0 þ β1 x1 þ β2 x2 þ β11 x1 2 þ β22 x2 2 þ β 12 x1 x2
416 Food Bioprocess Technol (2017) 10:413–433

Fig. 1 Plot of residual fit of


regression model for response. a
Appropriate residual fit. b
Inappropriate residual fit

The optimum point is found by calculating Δy/Δx1 and each individual response. The desirability function ap-
Δy/Δx2 and setting zero: proach is one of the most widely used methods in industry
for the optimization of multiple-response processes. It is
Δy=Δx1 ¼ β1 þ 2β 11 x1 þ β12 x2 ¼ 0 based on the idea that the Bquality^ of a product or pro-
Δy=Δx2 ¼ β2 þ 2β 22 x2 þ β12 x1 ¼ 0 cess which has multiple quality characteristics, with one
of them outside of some Bdesired^ limits, is completely
The above equations are solved to determine the values of unacceptable. The method finds operating conditions of X
x1 and x2. These values reveal the coded value of the factors, that provide the Bmost desirable^ response values. For
which give the maximum or minimum response. For optimi- each response Yi (x), a desirability function di (Yi) assigns
zation of process, the target point has to be found in the ranges numbers between 0 and 1 to the possible values of Yi,
of studied factors. with di (Yi) = 0 representing a completely undesirable
Generally, in the food industry, researchers are interest- value of Yi and di (Yi) = 1 representing a completely de-
ed in optimization of several responses concurrently, sirable or ideal response value. The individual desirabili-
which is complicated compared to optimization of pro- ties are then combined using the geometric mean, which
cesses with only one response. For this purpose, a desir- gives the overall desirability (D).
ability function is used which is a multi-criteria method-
ology. This methodology makes a desirability function for D ¼ ðd 1 ðY 1 Þ d 2 ðY 2 Þ…d k ðY k ÞÞ1=k
Food Bioprocess Technol (2017) 10:413–433 417

Fig. 2 Response surface (left)


and contour (right) plots of the
predicted model equation

where k represents the number of responses. It should be good efficiency in the modeling of quadratic functions and
noted that if any response Yi is completely undesirable, when the number of factors is higher than 2. Because in this
then the overall desirability is zero. case, a large number of experimental runs is required than can
Applying desirability function in food processes has usually be provided in practice and compared to the designs
several advantages such as economy, efficiency, and ob- that apply a smaller number of experimental runs such as CCD
jectivity in optimizing the multiple-responses processes and Box-Behnken design. If the number of design variables
(Bezerra et al., 2008). becomes large, a fraction of a full factorial design can be used
Measurement of response under the predicted optimum at the cost of estimating only a few combinations among var-
conditions and comparison of the observed responses with iables. This is called fractional factorial design and is usually
the responses predicted through the model is another impor- used for screening important design variables. For a 3n facto-
tant point. The difference between observed and predicted rial design, a (1/3)p fraction can be constructed, resulting in
responses must be as low as possible; this represents the pre- 3n–p points (Montgomery, 2008). FFD is appropriate to obtain
dicted optimal conditions are valid (Myers et al., 2016). a lot of information about the main effects in a proportionately
few number of runs. However, FFD is not appropriate to eval-
uate the interaction between factors due to deficiency of this
Design of Experiments in RSM design to provide information about interactions. The chroma-
tography and isolation applications are the main fields of FFD
An important section of RSM is the design of experiments that (Bezerra et al., 2008; Myers et al., 2016). Figure 3a shows a
is commonly abbreviated as DoE. The purpose of DoE is the FFD based on the study of all variables in three levels.
choice of the points where the response should be studied. A
comparison of some popular DoE methodologies is given in Box-Behnken Design
the following section.
The design that applies three levels (−1, 0, +1) for each factor
Full Factorial Design for the first time was designed by Box and Behnken (1960).
The Box-Behnken design (BBD) comprises a specific subset
Full factorial design has the appropriate efficiency and is com- of the factorial combinations from the 3k factorial design. In
parable with other designs such as central composite design addition, in a BBD, the experimental points are situated on a
(CCD). However, the full factorial design (FFD) has not a hypersphere equally distant from the central point (Fig. 3b).
418 Food Bioprocess Technol (2017) 10:413–433

Fig. 3 Experimental designs


based on the study of all variables
in three levels. a Full factorial
design (FFD). b Box-Behnken
design (BBD). c Central
composite design (CCD)

Although BBD is not rotatable, like CCD, some of its designs reasonable amount of information for testing lack-of-fit while
could be rotatable. Applying this design is popular in food not involving an unusually large number of experimental runs.
processes due to its economical design. BBD is appropriate Also, CCD is appropriate to study factors with three and/or
to evaluate interaction between factors and especially to study five levels. However, a CCD considers extreme points, which
processes without extreme points (where high levels of factors is not advisable for special processes such as extraction of a
involved in the process is difficult to implement, such as high compound sensitive to high temperature and pressure (Myers
temperature and pressure next to each other). In addition, et al., 2016). Experimental runs in CCD can be estimated by:
BBD is not appropriate to study factors with more than three
levels. In other words, only triplex levels (−1, 0, +1) are ap- N ¼ k 2 þ 2k þ cp
plied (Myers et al., 2016). The number of BBD experiments
can be calculated by the following equation: where k is the number of independent variables and cp is the
replicate number of the central point.
N ¼ 2k ðk−1Þ þ cp α values can be calculated by α = 2(k − p)/4; only quintuplet
levels are applied for the all factors (−α, −1, 0, +1, +α). α
where k is the number of factors and cp is the number of the value could be equal to 1 (face centered), which means this
central points (Fig. 3b). design has only three levels and the design space is then a cube
rather than a sphere (Myers et al., 2016).
Central Composite Design Table 1 shows the coded values of experimental matrix for
using FFD, BBD, and CCD designs. As it is obvious, the
A second-order model can be constructed efficiently with number of experiments in BBD is lower than FFD and
CCD. Box and Wilson (1951) developed this approach for CCD, 13 in the case of BBD compared to 27 and 15 for
the first time. The CCD (Fig. 3c) consists of three point types: FFD and CCD, respectively.
(1) factorial points, (2) a central point, and (3) axial points It should be acknowledged that software programs such
which are at a distance α from the central point. CCD is as Design Expert, Minitab, and SPSS greatly facilitate de-
appropriate for sequential experimentation and provides a sign and modeling by RSM. They typically normalize the
Food Bioprocess Technol (2017) 10:413–433 419

Table 1 Some experimental


matrices for RSM designs based Experiment no. Full factorial design Box-Behnken design Central composite
on three level variables (FFD) (BBD) design (CCD)

X1 X2 X3 X1 X2 X3 X1 X2 X3

1 −1 0 0 −1 −1 0 −1 −1 −1 Factorial points
2 −1 0 −1 1 −1 0 1 −1 −1
3 −1 −1 −1 −1 1 0 −1 1 −1
4 −1 −1 0 1 1 0 1 1 −1
5 −1 0 1 −1 0 −1 −1 −1 1
6 −1 1 0 1 0 −1 1 −1 1
7 −1 1 1 −1 0 1 −1 1 1
8 −1 −1 1 1 0 1 1 1 1
9 −1 1 −1 0 −1 −1 −α 0 0 Axial points
10 0 0 0 0 1 −1 α 0 0
11 0 0 −1 0 −1 1 0 −α 0
12 0 −1 −1 0 1 1 0 α 0
13 0 −1 0 0 0 0 0 0 −α
14 0 0 1 0 0 α
15 0 1 0 0 0 0 Central point
16 0 1 1
17 0 −1 1 0 0 0
18 0 1 −1 0 0 −1
19 1 0 0 0 −1 −1
20 1 0 −1 0 −1 0
21 1 −1 −1 0 0 1
22 1 −1 0 0 1 0
23 1 0 1 0 1 1
24 1 1 0 0 −1 1
25 1 1 1 0 1 −1
26 1 −1 1
27 1 1 −1

factors before performing the regression modeling, readily applying RSM in order to optimize the process conditions.
create graphical output so that the modeling results are There are several techniques for extraction of various bioac-
easily understood, allow for the creation of desirability tive compounds such as maceration, ultrasonic-assisted ex-
functions, and offer easy transformation of the responses traction (UAE), microwave-assisted extraction (MAE), super-
if needed. critical fluid extraction, and enzymatic extraction (Sarfarazi
et al., 2015; Khazaei et al., 2016; Wang, et al., 2012). In these
techniques, several factors including time, pH, temperature,
Applications of RSM in the Food Industry reagents concentration, irradiation time, flow rate, and solvent
to solid ratio affect extraction yield and process efficiency
In this section, we have reviewed some RSM studies pub- which can be considered as independent variables in RSM.
lished in different fields of food industry processes during We have classified the recent extraction studies into 12
the last few years. Generally, these processes involve CCD groups based on target compound and each group involves
and BBD designs. different independent variables and specific objectives; stud-
ies of phenolic compounds and polysaccharides comprise the
Optimization of the Extraction Processes major works in this field (Table 2). Maximizing the extraction
yield (MEY) has been the main target for these studies.
In recent years, numerous studies have been published in the Quality of analyte (target compound) is also important in ad-
literature about extraction of oil, protein, phenolic com- dition to the yield, which RSM can introduce in the optimal
pounds, pigments, polysaccharides, hydrocolloids, etc. by conditions for multiple-response processes. Among
Table 2 Some applications of RSM on extraction of different ingredients in the food industry
420

Target Process Independent variables Design Number of Randomization RSM Modeling validity Validation of Ref.
compound replication at modeling criteria predicted
central point graphical optimum
Number (types) Levels presented R2 Residual Lack- conditions
analyses of-fit

Vitamin Vitamin B3 from pork 2 (formic acid and ammonium 3 BBD 5 Done Done A – – Done (Saccani et al.,
products formate) 2005)
Vitamin E from wheat germ 3 (pressure, temperature, and flow 3 CCD 3 Not done Done SA A A Done (Ge et al., 2002)
rate CO2)
Pigment Natural pigments from annatto 4 (temperature, time, duty cycle, 5 CCD 7 Done Done A A A Done (Yolmeh et al.,
seed and seed to solvent ratio) 2014)
Anthocyanin from purple 3 (temperature, time, and 3 BBD 5 Done Done A SA A Not done (Fan et al., 2008)
sweet potato solid-liquid ratio)
Anthocyanin from purple 3 (SO2 concentration, time, and 3 CCD 4 Not done Done – – – Not done (Gao & Mazza,
sunflower hulls hull size) 1996)
Chlorophylls from Phaseolus 3 (number of extractions, 3 – 2 Done Done A – – Not done (Cubas et al.,
vulgaris L. extraction, and homogenization 2008)
time)
Natural dye from petals of the 3 (time, temperature, and mass of 5 BBD 6 Not done Done A A A Not done (Sinha et al., 2012)
flame of forest petals)
Lycopene from tomato 5 (solvent/meal ratio, number of 5 CCD 6 Not done Done A A I Done (D. Kaur et al.,
processing waste skin extraction, temperature, particle 2008)
size, and time)
Pigments from prickly pear 3 (temperature, time, and mass of 3 BBD 5 Done Done A A – Done (Maran &
the fruit) Manikandan,
2012)
Phenolic Phenolic compounds from 4 (temperature, time, pH, and ratio) 3 CCD 3 Not done Done A – – Not done (S. Rodrigues
compounds coconut shell powder et al., 2008)
Phenolic compounds from 4 (microwave power, time, ethanol 3 BBD 5 Not done Done SA A A Done (Bai et al., 2010)
apple pomace concentration, and ratio of
solvent to raw material)
Bioactive compounds from 3 (buffer to solid ratio, temperature, 5 CCD 6 Not done Done A A A Done (Gan & Latiff,
Parika speciosa pod and time) 2011)
Phenolic compounds from 3 (combination of solvent 3 BBD 3 Done Done A – A Done (Yang et al., 2009)
bark of Phyllanthus concentration, time, and
emblica L. temperature)
Bioactive compounds from 3 (ethanol content, temperature, 5 CCD 6 Done Done A A A Not done (Gong et al., 2012)
marigold and time)
Phenolic compounds from 3 (ethanol content, pH, and time) 3 CCD 2 Not done Done A – – Not done (Mylonaki et al.,
olive (Olea europaea) leaf 2008)
Phenolic compounds from 3 (ethanol to water ratio, 3 BBD 3 Done Done A A A Not done (Juntachote et al.,
lemon grass, galangal, holy temperature, and time) 2006)
basil, and rosemary
Protein Gelatine from grass carp 4 (HCl concentration, pretreatment, 3 BBD – Not done Done A – A Not done (Kasankala et al.,
(Catenopharyngodon temperature, and time) 2007)
idella) fish skin
Collagen from grass carp 3 BBD 5 Done Done A A A Not done
Food Bioprocess Technol (2017) 10:413–433
Table 2 (continued)

Target Process Independent variables Design Number of Randomization RSM Modeling validity Validation of Ref.
compound replication at modeling criteria predicted
central point graphical optimum
Number (types) Levels presented R2 Residual Lack- conditions
analyses of-fit

3 (acetic-acid concentration, time, (L. Wang et al.,


and temperature) 2008)
Protein from pumpkin seeds 3 (liquid: solid ratio, NaCl 3 CCD 3 Not done Done A – – Done (Tabaraki &
concentration, and time) Nateghi, 2011)
Antioxidants Natural antioxidants from rice 3 (ethanol concentration, 3 CCD 2 Done Done A – A Done (Pompeu et al.,
bran temperature, and time) 2009)
Phenolic antioxidants from 3 (ethanol proportion, HCl 5 CCD 3 Not done Done A Done (Wijngaard &
Food Bioprocess Technol (2017) 10:413–433

A –
fruits of Euterpe oleracea concentration, and temperature) Brunton, 2010)
Antioxidants from apple 3 (ethanol concentration, 5 CCD 6 Done Done A A A Not done (H. Sin et al.,
pomace temperature, and time) 2006a, b)
Juice Sapodilla juice from sapodilla 2 (temperature, and time) 3 CCD 4 Done Done A – I Done (Lee et al., 2006)
Banana juice from banana 2 (temperature and time) 3 CCD 5 Done Done A – A Not done (Zhong & Wang,
2010)
Polysaccharide Polysaccharide from dried 3 (ultrasonic power, ratio of water 3 BBD 5 Not done Done A – A Done (Gan et al., 2010)
longan pulp to material, and time)
Alcohol insoluble 2 (temperature and time) 5 CCD 5 Not done Done A A – Done (Aliakbarian et al.,
polysaccharides from 2008; J. H. Xie
Parkia speciosa pod et al., 2010)
Polysaccharide from 3 (temperature, time, and 3 CCD 3 Not done Done A – A Done (Xie et al., 2012)
Cyclocarya paliurus water/material ratio)
(Batal.) Iljinskaja
Polysaccharide from waste of 2 (temperature and time) 5 CCD 5 Not done Done A A A Not done (Yoshida et al.,
corn starch 2010)
Oil Oil from olive 2 (malaxation time and enzyme 3 FFD 3 Not done Done A A I Not done Aliakbarian et al.,
concentration) 2008)
Oil from pomegranate seed oil 3 (pressure, temperature and CO2 5 CCD 3 Done Done A – – Not done (Liu et al., 2003)
flow rate)
Hydrocolloids Hydrocolloid from cress seed 4 (water: seed, pH, temperature, 3 CCD 8 Done Done A A A Done (Karazhiyan et al.,
and time) 2011)
Flavonoid Flavonoid from Spearmint 3 (temperature, pressure, and 3 CCD 6 Done Done A – A Not done (Bimakr et al.,
leaves co-solvent flow rate) 2012)
Essential oil Rosemary essential oil from 3 (processing pressure, moisture 5 CCD 8 Not done Done A – A Not done (Rezzoug et al.,
rosemary leave content, and processing time) 2005)
Essential oil from rhizomes of 4 (temperature, pressure, time, and 5 CCD 6 Not done Done A – I Not done ( Wang et al.,
Cyperus rotundus Linn Co2 flow rate) 2012)
Pectin Pectin from lemon by-product 3 (temperature, pH, and time) 3 CCD 1 Not done Done A A A Not done (Masmoudi et al.,
2008)
Pectin acid from passion fruit 3 (temperature, pH, and time) 5 CCD 1 Not done Done A A A Not done (Masmoudi et al.,
peel 2008)
421

A appropriate, I inappropriate, SA semi-appropriate; −: not studied, FFD full factorial design, BBD Box-Behnken design, CCD central composite design
422 Food Bioprocess Technol (2017) 10:413–433

independent variables studied in optimization of extraction higher number of papers considering drying optimization
processes, temperature, time, pH, and solid to solvent ratio have implemented validation of predicted optimum condi-
were most widely used. Also, the CCD was the major design tions. Among papers reviewed in Table 3, Madamba (2002)
in the extraction studies by RSM. Randomization of experi- has employed RSM appropriately because of the randomiza-
ments and replication on central point are important criteria tion, RSM modeling graphical presentation, modeling validity
that should be used in RSM designs to validate RSM model- criteria, and validation of predicted optimum conditions
ing; however, it was not observed in some published papers, adequately.
as shown in Table 2. Validation of predicted optimum condi-
tions is another important issue when applying the RSM ap- Application of RSM in the Formulation Processes
proach, which was not seen in some of these studies. Among
papers reviewed in Table 2, Sarfarazi et al. (2015), Khazaei Since several factors are considered in formulation of food
et al. (2016), Yolmeh et al. (2014), Maran and Manikandan products, researcher are seeking for optimization of each re-
(2012), Yang et al. (2009), Gan et al. (2010), and Karazhiyan sponse (factor) which could be performed by RSM, as this
et al. (2011) employed RSM appropriately since they have approach is useful for multi-response processes. RSM has
applied the randomization, RSM modeling graphical presen- been used for optimizing formulations of different foods such
tation, modeling validity criteria, and validation of predicted as functional short dough biscuits (Gallagher, et al., 2003),
optimum conditions adequately in their studies. However, soy-coffee beverage (Felberg, et al., 2010), low-calorie mixed
RSM has been used inappropriately in the rest of them. fruit jelly (Acosta, et al., 2008), extruded snack food (Thakur
Most of the studies reviewed in Table 2 revealed that RSM & Saxena, 2000), cream (Moulai Mostefa et al., 2006), gluten-
was successfully applied to optimize extraction conditions by free bread (McCarthy, et al., 2005; Sanchez, et al., 2004), pork
different techniques, because of good R2-adj and R2-pred frankfurters (Pappa, et al., 2000), walnut oil-in-water beverage
values, insignificant P values for lack-of-fit value, and good emulsion (Gharibzahedi, et al., 2012), Iranian white brine
compatibility of the predicted and experimental values. cheese (Alizadeh, et al., 2005), homogenized infant foods
Therefore, the RSM algorithm could accurately estimate the (Martinez, et al., 2004), and sweet potato-based pasta
results. (Singh, et al., 2004). Gharibzahedi et al. (2012) studied the
effects of Arabic gum content and walnut-oil concentration on
Optimization of the Drying Processes properties of prepared emulsions, namely turbidity loss rate,
density, particle size, and stability, using RSM. They reported
Drying is one of the techniques to improve the shelf life of that the optimum formulation was containing 3% (w/w) wal-
foods such as fruits and vegetables which have a large mois- nut oil and 9.62% (w/w) Arabic gum.
ture content (Kumar, et al., 2014). There are various tech- The above studies successfully used RSM to optimize for-
niques for drying and dehydration of foods such as hot air mulation processes revealed by suitable validity criteria (good
drying, osmotic dehydration, microwave-vacuum drying, flu- R2-adj and R2-pred values, insignificant P values for lack-of-
idized bed drying, IR-assisted freeze drying, and microwave- fit, and good compatibility of the predicted and experimental
assisted hot air drying (Madamba, 2002). Producing a dried values). In these studies that applied RSM to optimize formu-
product with good quality is dependent on several factors such lation, different factors as independent variable were investi-
as temperature, time, thickness, air speed, vacuum rate, micro- gated, which are dependent on case to case; for instance, in the
wave power, concentration of osmotic solution, etc. (Burande, work of Gharibzahedi et al. (2012), concentrations of hydro-
et al., 2008). Thus, optimization of these process conditions is colloid, oil, sweetener, and conditions of process such as tem-
important in terms of improving the quality and minimizing perature and pH were the main independent variables. On the
the process costs. other hand, physicochemical, textural, and rheological prop-
Statistical approaches such as RSM can introduce the op- erties were the main dependent variables.
timal conditions for the drying process based on several pro-
cess responses such as water loss, solid gain, final moisture, Application of RSM in Microencapsulation
color, and rehydration ratio. Table 3 shows different studies and Controlled-Release Processes
which have applied RSM to optimize drying conditions cate-
gorized by the type of drying technique. Temperature, time, Microencapsulation is a technique in which very small parti-
osmotic solution concentration, and the product thickness cles or droplets are surrounded by a wall material to give small
have been widely used as independent variables in these op- capsules. The technique is widely used in the food and drug
timizing studies on drying processes. As it is obvious in industry to protect bioactives from environmental deteriora-
Table 3, in most of these studies, CCD has been applied in tive factors and also to develop and produce improved food
order to investigate and optimize the drying conditions. Also, and drug delivery systems. However, there are many factors
compared with the previous section (extraction processes), influencing the stability and functionality of the
Table 3 Some applications of RSM on different drying processes in the food industry

Drying method Sample Independent variables Design Number of Randomization RSM Modeling validity Validation of predicted Ref.
replication at modeling criteria optimum conditions
Number (types) Levels central point graphical
presented R2 Residual Lack-
analyses of-fit

Hot air drying Olive leaves 3 (temperature, air velocity, and 5 CCD 6 Not done Done A A A Not done (Erbay & Icier,
time) 2009)
Agricultural (temperature, thickness, vacuum 3 BBD 3 Done Done A A A Done (Kumar et al.,
crops pressure, air speed, and RH) 2014;
Madamba,
2002)
Food Bioprocess Technol (2017) 10:413–433

Osmotic Potato 4 (temperature, sucrose 5 CCD 7 Not done Done A A A Not done (Eren &
dehydration concentration, salt Kaymak-Erte-
concentration, and time) kin, 2007)
Kiwifruit 4 (sucrose concentration, slice 5 CCD 7 Not done Done A – A Not done (Cao et al., 2006)
thickness, temperature, and
time)
Carrot cubes 3 (time, temperature, and NaCl 5 CCD 6 Done Done A A A Not done (Bahadur Singh
concentration) et al., 2007)
Aonla slices 4 (temperature, solution to fruit 3 CCD 3 Done Done A A A Not done (Alam et al.,
ratio (STFR), sugar 2010)
concentration, and time)
Diced green 4 (salt, sorbitol, temperature, and 5 CCD 6 Done Done A A A Not done (Ozdemir et al.,
peppers time) 2008)
Red paprika 2 (sucrose concentration, and 5 CCD 5 Not done Done A – A Done (Ade-Omowaye
(Capsicum sodium chloride et al., 2002)
annuum L.) concentration)
Microwave-vacuum Apple slices 3 (microwave power, vacuum 5 CCD 9 Not done Done A A A Done (Han et al., 2010)
drying level, and initial moisture
content)
Button 3 (microwave power, system 5 CCD 6 Not done Done A – A Not done (Han et al., 2010)
mushrooms pressure, and thickness)
(Agaricus
bisporous)
Fluidized bed drying Green peas 4 (temperature, tempering time, 5 CCD 7 Not done Done A – – Not done (Burande et al.,
pretreatment, and mass 2008)
per unit area)
FIR-assisted freeze Yam slices 3 (temperature, distance, and 3 BBD 3 Not done Done A A A Done (Lin et al., 2007)
thickness)
Microwave-assisted Okra 3 (air velocity, air temperature, 5 CCD 6 Not done Done A – A Not done (Kumar et al.,
hot air drying and MW power) 2014)

A Appropriate, I inappropriate, SA semi-appropriate, FFD full factorial design, BBD Box-Behnken design, CCD central composite design
423
424 Food Bioprocess Technol (2017) 10:413–433

microcapsules, such as the type of wall material, the properties that the optimum rate of drug release was observed at 1.8%
of the core materials (volatility and concentration), the char- chitosan concentration, a pH 8.7 for the tripolyphosphate
acteristics of the initial emulsion (droplets size, total solids, solution and 9.7 min cross-linking time. In addition to these
and viscosity), and the conditions of the drying process (tem- studies, controlled-release of AZT-loaded microspheres,
perature, air flow, and humidity) (Jafari, et al., 2008). Thus, it nicardipine hydrochloride, and mucoadhesive tablets of
is important to optimize the microencapsulation process by atenolol have been optimized using RSM by Abu-Izza
techniques such as RSM. et al. (1996), Huang et al. (2005), and Singh et al. (2006),
In the recent years, some researchers have applied RSM respectively.
to optimize the parameters of the microencapsulation pro-
cess efficiently for oils, probiotics, vitamins, minerals, and Optimization of the Enzymatic Hydrolysis
controlled-release of tablets. For instance, Chen et al. and Clarification
(2006) developed an optimal composition for encapsulat-
ing probiotics with sodium alginate blended with peptides, The enzymatic hydrolysis depends on many factors such as
fructooligosaccharides, and isomaltooligosaccharides as type and specificity of the enzyme, hydrolysis time, enzyme
the wall materials through the RSM model. They reported concentration, incubation temperature, and pH. Since the
that optimal combination of wall materials for the micro- enzymatic hydrolysis is widely used in the food industry,
capsules was found to be 3% sodium alginate mixed with it is necessary to investigate enzymatic performance during
1% peptides, 3% fructooligosaccharides, and 0% hydrolysis and optimize the process for obtaining a high
isomaltooligosaccharides. In another study, these research- yield by utilizing low-enzyme contents. Classical tech-
er reported that 1% sodium alginate mixed with 1% peptide niques are laborious, time consuming, and often do not
and 3% fructooligosaccharides as wall materials would guarantee the determination of optimal conditions
provide the highest protection for the probiotics (Kurozawa, et al., 2008). Recently, RSM has been applied
(Bifidobacterium bifidum, Bifidobacterium longum, and, for the optimization of enzymatic hydrolysis processes. For
Lactobacillus acidophilus) (Chen, et al., 2005). instance, Kurozawa et al. (2008) studied optimization of the
Results of Kha et al. (2014), which optimized microencap- enzymatic hydrolysis of chicken meat using RSM, and their
sulation of gac oil using spray drying by RSM, revealed that a results revealed that a condition with 52.5 °C, 4.2% (w/w)
combination with wall concentration of 29.5% and oil load of enzyme/substrate ratio, and pH of 8.00 was found as the
0.2 was the optimum condition for the microencapsulation. optimum conditions. In another study, Pericin et al.
Huynh et al. (2008) introduced a condition with 40% feed (2009) reported 0.137% (v/v) enzyme concentration,
concentration (w/w), 18% oil concentration, and 65 °C outlet 0.84% (w/v) NaCl concentration, and 32.5 h as the optimum
air temperature as optimum conditions of the microencapsu- conditions for enzymatic hydrolysis of protein isolate from
lation. Similarly, microencapsulation of rice bran oil, sunflow- hull-less pumpkin oil cake. Results of Kaur et al. (2009)
er oil, and flaxseed oil has been optimized using RSM by Suh showed that a condition with the enzyme concentration
et al. (2007), Ahn et al. (2008), and Tonon et al. (2011), re- 0.70 mg/100 g guava pulp, 7.27 h incubation, and 43.3 °C
spectively. Recently, Assadpour et al. (2016) applied RSM for is ideal for enzymatic hydrolysis pretreatment of juice re-
nano-emulsification and encapsulation of folic acid and Salimi covery from guava fruit. Similarly, RSM has been used to
et al. (2015) encapsulated lycopene through emulsification optimize the enzymatic hydrolysis condition of dogfish
optimization by RSM. (Squalus acanthias) muscle (Schwabe & Büllesbach,
Pharmaceutical formulators are also seeking to find the 2013), maize starch (Kunamneni & Singh, 2005), whey
right combination of pellets, which will produce a product protein isolate (Cheison, et al., 2007), and pectin
with optimum properties. Optimizing formulation is very (Rodríguez-Nogales, et al., 2007).
important when it is a controlled-release dosage form, be- The turbidity and viscosity of fruit juice are caused mainly
cause many factors can affect the release rate. RSM has by the polysaccharides such as pectin, cellulose, hemicellu-
been used to optimize formulation of drugs in the develop- lose, lignin, and starch. These colloids may lead to the fouling
ment of a product with controlled-release property. For ex- problem during the filtration process, and an enzymatic treat-
ample, Kim et al. (2007) studied formulation and optimiza- ment is necessary to prevent this problem with enzymes such
tion of a novel oral controlled delivery system for as pectinase and cellulose. Generally, these enzymes increase
tamsulosin hydrochloride using RSM. They found combi- juice yield, soluble solids content, and clarity in fruit and
nation of 10% hydroxypropyl methylcellulose and 10% hy- vegetable juices (Alvarez & Canet, 1999). A number of fac-
droxypropyl methylcellulose phthalate at a coating level of tors could affect the enzymatic clarification such as enzyme
25% as optimal coating formulation. In another study, Ko concentration, time, and temperature incubation of the
et al. (2003) developed and optimized chitosan microparti- treatment. Usually, in these studies, the responses are
cles for controlled drug release using RSM. They reported turbidity, clarity, viscosity, and antioxidant activity. RSM has
Food Bioprocess Technol (2017) 10:413–433 425

been used successfully to optimize enzymatic clarification 10% lower when the potato cuts were blanched in water that
processes. For instance, Rai et al. (2004) found 99.27 min was previously used for blanching, leading to over 10% less
time, 41.89 °C temperature, and 0.0004% (w/v) concentration reduction in the final acrylamide content too. Yolmeh and
of pectinase as the best conditions for clarification of mosambi Najafzadeh (2014) investigated the ultrasound blanching of
juice using CCD. Lee et al. (2006) reported that 0.084% green bean by RSM and they found that a temperature of
pectinase concentration, incubation temperature of 43.2 °C, 90 °C, sonication time of 58.27 s, and duty cycle of 0.79 s
and incubation time of 80 min are the best conditions for were optimum conditions in which residual peroxidase activ-
clarifying banana juice. The results of Chen et al. (2012) re- ity and vitamin C loss were measured as 9.64 and 8.92%,
vealed that the optimal enzymatic clarification of green aspar- respectively. In addition to these examples, other studies have
agus juice using CCD is pectinase concentration of 1.45%, also been carried out to optimize blanching of frozen Jalapeno
incubation temperature of 40.56 °C and pH of 4.43. In some pepper (Capsicum annuum), potato (Solanum tuberosum L),
other published works, RSM has been used for optimizing frozen/thawed mashed potatoes, and olive leaf through RSM
enzymatic clarification of sapodilla juice (Sin et al., 2006a, by Quintero-Ramos et al. (1998), Reyes-Moreno et al. (2002),
b), litchis (Litchi Chinensis Sonn.) juice (Sanchez et al., Fernandez et al. (2006), and Stamatopoulos et al. (2012),
2004), white pitaya juice (Nur‘Aliaa, et al., 2010), and pome- respectively.
granate juice (Neifar et al., 2011).
Application of RSM in Production of Microbial Enzymes
Optimization of the Blanching Process and Other Metabolites

Blanching is a heat treatment which is widely used in the agro- Production of microbial enzymes and metabolites in microor-
food sector and particularly important in the processing of ganisms is highly affected by several factors, such as type of
green vegetables to inactivate the enzymes involved in the carbon and nitrogen sources, temperature, pH, metal ions,
spoilage of fresh vegetables (Garrote, et al., 2004). There are incubation time, and inoculum volume (Açıkel, et al., 2010;
different techniques for blanching such as steam or hot water, Liew, et al., 2005). Therefore, it is significant to optimize the
microwave, and ultrasound. Several factors are important dur- production conditions of microbial enzymes and metabolites,
ing blanching, including temperature, time, thickness, particle which can be carried out by conventional and statistical tech-
size, ultrasound and microwave power, and concentration of niques. Statistical experimental approaches such as RSM are
solvents, which should be optimized in order to achieve an much better since they suggest simultaneous, systematic, and
appropriate blanching process. RSM has been used to opti- efficient variation of all the components for process optimiza-
mize different blanching processes by considering parameters tion economically (Abdel-Fattah, 2002).
such as enzymatic activity (EA), vitamin C content, color, etc. As shown in Table 4, the highest EA of xylanase was mea-
as dependent variables or responses (Yolmeh & Najafzadeh, sured at 61 U/mL from Chaetomium thermophilum by
2014). As an example, Jackson et al. (1996) studied optimi- Katapodis et al. (2007). They reported that this EA was ob-
zation of blanching for banana chips using RSM and reported served in 3.9% (w/v) wheat straw and 0.7% (w/v) sodium
that the crispiest chips could be produced at blanching condi- nitrate. The highest EA of lipase was observed at 6230 IU/
tions of 69 °C and 22 min. Alvarez and Canet (1999) studied mL from Candida sp. 99–125 by He and Tan (2006) at 4.18%
optimization of stepwise blanching of frozen-thawed potato (w/v) soybean oil, 5.84% soybean powder, 0.284% K2HPO4,
tissues and showed that ranges of temperature of 60–65 °C 0.1% KH2PO4, 0.1% (NH4)2SO4, 0.05% MgSO4, and 0.1%
and time of 25–35 min were optimum conditions for the Span 60 as the optimum culture medium. The highest EA of
blanching. Ismail and Revathi (2006) evaluated the effects α-amylase was measured at 6583 U/g from Aspergillus oryzae
of blanching time, evaporation time, temperature, and hydro- by Francis et al. (2003) at 30 °C, an initial moisture of 70%,
colloid on physical properties of chili (Capsicum annum var and an inoculum rate of 1 × 107 spores/g dry substrate as the
kulai) puree by RSM. They reported that a complete inactiva- optimum conditions. The highest EA of phytase was observed
tion of the enzymes was obtained with the blanching temper- at 2250 U/L from Escherichia coli by Sunitha et al. (1999) at
ature of 100 °C for 6 min. First-order reaction kinetics fitted 11.75 g/L trptone, yeast extract 5.88 g/L, and 5.32 g/L NaCl as
adequately to predict color loss and pectin in order to improve the optimum condition. The highest EA of protease was
the viscosity and total soluble solids of the puree. measured at 1939 U/mL from Bacillus sp. by Puri et al.
Mestdagh et al. (2008) used a CCD to investigate the effect (2002) at 15 mg/mL starch, 7.5 mg/mL peptone, 144 h
of blanching time and temperature on the extraction of reduc- incubation time, and 1% inoculums density as the optimum
ing sugars from potato strips and slices. Their results revealed condition.
that blanching at 70 °C for 10 min was more efficient than It is clear in Table 4 that glucose, oil, and yeast extract were
lower temperatures that appeared more time consuming. used most widely as independent variables in optimizing pro-
However, the extraction yield of reducing sugars was over duction studies of the microbial enzymes and some other
Table 4 Some applications of RSM on production of microbial enzymes and some other metabolites in the food industry
426

Enzyme or Microorganism Independent variables Design Number of Randomization RSM modeling Modeling validity Validation of predicted Ref.
metabolite replication at graphical criteria optimum conditions
Number (types) Levels central point presented
R2 Residual Lack-
analyses of-fit

Xylanase Bacillus circulans 3 (xylan, pH, and cultivation time) 3 FFD 1 Not done Done A – – Not done (Bocchini et al.,
2002)
Xylanase Chaetomium 2 (NaNO3, and wheat straw) 5 CCD 3 Not done Done A – – Done (Katapodis et al.,
thermophilum 2007)
Xylanase Streptomyces sp. 3 (wheat bran, KNO3, and xylose) 5 CCD 5 Done Done A A A Done (Coman &
P12-137 Bahrim, 2011)
Lipase Rhizopus delemar 5 (sucrose, molasses sucrose, yeast 3 BBD 6 Done Done A A A Not done (Açıkel et al.,
extract, sunflower oil, and 2010)
Tween 80)
Lipase Aspergillus carneus 5 (sunflower oil, peptone, glucose, 3 CCD 7 Done Done A A A Done (Kaushik et al.,
agitation rate, and inoculum 2006)
period)
Lipase B. subtilis PTCC 1720 4 (temperature, sucrose, peptone, 5 CCD 5 Done Done A A A Done (Esmaeili et al.,
and corn oil) 2015)
Lipase Candida sp. 99-125 3 (Soybean oil, soybean powder, and 5 CCD 6 Not done Done A – – Done (He & Tan, 2006)
K2HPO4)
Lipase Candida cylindracea 5 (yeast extract, malt extract, 5 CCD 3 Done Done A – – Done (Muralidhar
peptone, glucose, and Tween 80) et al., 2001)
α-amylase Aspergillus oryzae 4 (medium, temperature, moisture 3 BBD 3 Not done Done A – – Done (Francis et al.,
content, and number of spores) 2003)
α-amylase Bacillus sp. 4 (starch, glycerol, YE, and 5 CCD 6 Not done Done A A A Done (Tanyildizi et al.,
peptone) 2005)
α-amylase B. amyloliquefaciens 3 (substrate, incubation period, and 3 BBD 6 Done Done A A A Done (Gangadharan
CaCl2) et al., 2008)
Phytase Pichia anomala 3 (glucose, beef extract, and 5 CCD 8 Not done Done A – I Done (Vohra &
inoculum size) Satyanarayan-
a, 2002)
Phytase Escherichia coli 3 (tryptone, yeast extract, and 5 CCD 4 Done Done A – – Not done (Sunitha et al.,
NaCl) 1999)
Phytase Kodamaea ohmeri BG3 3 (oats, ammonium sulfate, and 5 CCD 6 Not done Done A – – Done (Li et al., 2008)
pH)
Protease B. subtilis PTCC 1720 4 (temperature, sucrose, peptone, 5 CCD 5 Done Done A A A Done (Esmaeili et al.,
and corn oil) 2015)
Alkaline B. mojavensis 5 (casamino acids, glucose, 3 CCD 8 Done Done A – – Not done (Beg et al., 2003)
protease inoculum age, incubation time,
and agitation)
Protease Pseudomonas sp. 3 (pH, temperature, and inoculum 5 CCD 1 Not done Done A – – Done (Dutta et al.,
volume) 2004)
Alkaline Bacillus sp. 4 (starch, peptone, inoculum 3 CCD 6 Not done Done A – – Done (Puri et al., 2002)
protease density, and incubation time)
Aspergillus niger 2 (agitation, and aeration) 5 CCD 5 Done Done A – – Not done (Liu et al., 2003)
Food Bioprocess Technol (2017) 10:413–433
Table 4 (continued)

Enzyme or Microorganism Independent variables Design Number of Randomization RSM modeling Modeling validity Validation of predicted Ref.
metabolite replication at graphical criteria optimum conditions
Number (types) Levels central point presented
R2 Residual Lack-
analyses of-fit

Glucose
oxidase
Pectinase Kluyveromyces 3 (temperature, pH, and time) 5 CCD 6 Done Done A A A Not done (Moyo et al.,
wickerhamii 2003)
Nattokinase B. subtilis 4 (glucose, peptone, magnesium 5 CCD 5 Not done Done A A I Not done (Deepak et al.,
sulfate, and calcium chloride) 2008)
Inulinase Kluyveromyces S120 3 (inulin, corn steep liquor, and 3 BBD 3 Not done Done Not done (Xiong et al.,
Food Bioprocess Technol (2017) 10:413–433

A – –
(NH4)2SO4) 2007)
Elastase Bacillus sp. EL31410 6 (glucose, casein, corn steep flour, 6 CCD 3 Not done Done A – – Not done (Q. Chen et al.,
K2HPO4, MgSO4O7H2, and pH) 2002)
β glucanase Bacillus subtilis 6 (barley flour, corn flour, soybean 3 CCD 4 Not done Done A – – Done (Tang et al.,
ZJF-1A5 flour, KH2PO4, MgSO4.7H2O, 2004)
and CaCl2)
Hydrogen Ethanoligenens 3 (glucose, Fe2+, and Mg2+) 3 BBD 3 Done Done A – – Done (Guo et al.,
harbinense B49 2009)
Ethanol Sago starch 3 (temperature, time, and pH) 5 CCD 6 Done Done A – – Not done (Ratnam et al.,
2003)
– Recombinant 5 (medium type, glucose, 5 CCD 5 Not done Done A A – Done (Nikerel et al.,
Escherichia coli (NH4)2HPO4, KH2PO4, and 2006)
Mg SO4)
– Lactobacillus rhamnosus 4 (pH, vitamin solution, glucose, 5 CCD 6 Done Done A – A Done (Liew et al.,
and yeast extract) 2005)
Biosurfactant Lactococcus lactis 53 6 (peptone, meat extract, yeast 3 FFD 3 Done Done A A – Not done (L. Rodrigues
and Streptococcus extract, lactose, ammonium et al., 2006)
thermophilus A citrate, and KH2PO4)
Bacteriocin B. licheniformis strain 3 (temperature, pH, and cheese 5 CCD 3 Not done Done A – – Not done (Cladera-Olivera
P40 whey) et al., 2004)

A appropriate, I inappropriate, SA semi-appropriate, FFD full factorial design, BBD Box-Behnken design, CCD central composite design
427
428 Food Bioprocess Technol (2017) 10:413–433

metabolites. Among papers reviewed in Table 4, Coman and usually as systems of interconnected Bneurons^ which can
Bahrim (2011), Kaushik et al. (2006), Esmaeili et al. (2015), compute values from inputs, and are capable of machine
Gangadharan et al. (2008), and Liew et al. (2005) employed learning as well as pattern recognition giving to their adap-
RSM more appropriately. tive nature. This approach presents an interesting chance to
preparing non-linear modeling for response surface and op-
timization of food industry processes (Tran, et al., 2007).
Challenges of Applying RSM in Food Processes The ANN technique applies modeling of complex process-
es that is possible to study without complicated equations.
It was shown that there are much potential for applying RSM in The ANN methodology has several advantages compared
the food industry processes. However, this technique is not al- to regression models, such as more flexibility concerned
ways successful and there are some important points to consider with the number and type of experimental data, high pre-
when using it for achieving a successful performance including: dictive power, more accuracy, using all data, and not re-
quired to normalize data (Bezerra et al., 2008).
& Correct choice of the range of independent variables: In Table 5, some studies regarding the application of RSM
Enough preliminary work or experience is needed to select combined with ANN for optimization of food processes
the appropriate range of each factor, which directly affects (mainly isolation studies) has been presented. For example,
the success of RSM optimization. Sinha et al. (2013) used RSM and ANN to develop predictive
& Correct selection of the polynomial model: A second- models for simulation and optimization of the annatto dye
order equation is used commonly in RSM methodology. extraction process and they compared both modeling method-
If the trend of the responses in the studied range of factors ologies (RSM and ANN) statistically by R2, root mean square
is not suitable to depict with second-order equation, the error (RMSE), and AAD based on the validation data set.
range of independent variables or the form of the depen- Their results revealed that ANN had better prediction perfor-
dent should be transformed to a suitable form, in which the mance than RSM.
trend of the responses could be depicted with this equation In a study by Shahraki et al. (2014), they applied RSM
(Baş & Boyacı, 2007). The RSM approach needs to be combined with genetic algorithms (GA) to optimize the fluid-
regressed with a polynomial equation. In addition, the ized bed drying of sesame seeds and their results revealed that
number of terms in polynomial equation is limited to num- GA can improve the equation coefficients obtained by RSM.
ber of design points, as well as choice of the suitable
polynomial equation can be very laborious because each
response needs its own distinct polynomial equation. Application of RSM in Non-food Processes
Thus, the accuracy of the RSM modeling can be increased
through combining with other modeling techniques such In addition to food processes, RSM has been successfully
as artificial neural network (ANN). used to optimize non-food processes such as biochemical
and chemical engineering processes. Both BBD and CCD
ANNs present an option to the polynomial regression designs have been used in different non-food process studies,
method as a modeling approach. ANNs are presented for example in capillary zone electrophoresis (BBD, Suresh

Table 5 Some applications of RSM combined with ANNs in the food industry processes

Analyte Sample Analytical technique Objective of study Ref.

Natural dye Annatto seed MAE Development predictive models for simulation (Sinha et al., 2013)
and optimization of the dye extraction process
Herbicide Water HPLC Optimizing of the linear gradient separation of 10 (Tran et al., 2007)
herbicides consisting of a mixture of acids,
bases, and neutrals employing a single ANN
for modeling the response surface
Neuroprotective peptide Mixture of HPLC Sing experimental design conjunction with (Novotná et al., 2005)
peptide artificial neural networks for optimization of
isocratic ion-pair separation of neuroprotective
peptides
Cis- and trans- resveratrol Australian Capillary zone Optimizing the solid-phase extraction employing (Spanilá et al., 2005)
wine electrophoresis central composition design and ANN
Hydrochlorothiazide and Pharmaceutical HPLC Comparing artificial neural networks for (Agatonovic-Kustrin et al.,
amiloride response surface modeling in HPLC with 1998)
multiple regression methods
Food Bioprocess Technol (2017) 10:413–433 429

Babu et al., 2004), microwave-assisted extraction of persistent Alam, M. S., Amarjit, S., & Sawhney, B. (2010). Response surface opti-
mization of osmotic dehydration process for aonla slices. Journal of
organochlorine pesticides (BBD, Gfrerer and Lankmayr,
Food Science and Technology, 47(1), 47–54.
2005), separation of β-blockers by ion-pair capillary electro- Aliakbarian, B., De Faveri, D., Converti, A., & Perego, P. (2008).
phoresis (BBD, Servais et al., 2002), optimization of cadmium Optimisation of olive oil extraction by means of enzyme processing
determination by flame atomic absorption spectrometry aids using response surface methodology. Biochemical Engineering
Journal, 42(1), 34–40.
(BBD, Souza et al., 2005), biosensor development (CCD,
Alizadeh, M., Hamedi, M., & Khosroshahi, A. (2005). Optimizing sen-
Zhang et al., 2011; Mirmoghtadaie et al., 2013), optimization sorial quality of Iranian white brine cheese using response surface
extraction of human antibodies (CCD, Rosa et al., 2007), and methodology. Journal of Food Science, 70(4), S299–S303.
optimization extraction of five 1,4-dihydropyridines calcium Alvarez, M. D., & Canet, W. (1999). Optimization of stepwise blanching
channel antagonists from human plasma (CCD, Baranda et al., of frozen-thawed potato tissues (cv. Monalisa). European Food
Research and Technology, 210(2), 102–108.
2005). A wide range of dependent and independent variables
Assadpour, E., Maghsoudlou, Y., Jafari, S.-M., Ghorbani, M., & Aalami,
have been investigated in these processes, which are based on M. (2016). Optimization of folic acid nano-emulsification and en-
the case and target. capsulation by maltodextrin-whey protein double emulsions.
International Journal of Biological Macromolecules, 86, 197–207.
Bai, X. L., Yue, T. L., Yuan, Y. H., & Zhang, H. W. (2010). Optimization
of microwave-assisted extraction of polyphenols from apple pomace
Conclusions using response surface methodology and HPLC analysis. Journal of
Separation Science, 33(23–24), 3751–3758.
Nowadays, application of RSM in the optimization of differ- Baş, D., & Boyacı, İ. H. (2007). Modeling and optimization I: usability of
ent food industry processes is common due to its advantages response surface methodology. Journal of Food Engineering, 78(3),
836–845.
compared with conventional methods. By reviewing the liter- Beg, Q. K., Sahai, V., & Gupta, R. (2003). Statistical media optimization
ature, it was shown that CCD was the most utilized second- and alkaline protease production from Bacillus mojavensis in a bio-
order experimental design for optimizing the food processes. reactor. Process Biochemistry, 39(2), 203–209.
Also, the RSM approach was only usable for optimizing the Bezerra, M. A., Santelli, R. E., Oliveira, E. P., Villar, L. S., & Escaleira, L.
A. (2008). Response surface methodology (RSM) as a tool for op-
processes which can be depicted with a second-order polyno-
timization in analytical chemistry. Talanta, 76(5), 965–977.
mial equation. Moreover, it is important to use some criteria Bimakr, M., Rahman, R. A., Ganjloo, A., Taip, F. S., Salleh, L. M., &
such as lack-of-fit and AAD to assay the suitability of fitting Sarker, M. Z. I. (2012). Optimization of supercritical carbon dioxide
the response surface models, in addition to R2. Finally, an extraction of bioactive flavonoid compounds from spearmint
adaptive learning technique, which uses ANN merged with (Mentha spicata L.) leaves by using response surface methodology.
Food and Bioprocess Technology, 5(3), 912–920.
RSM, can be helpful to model the dependence relation and it Bocchini, D., Alves-Prado, H., Baida, L., Roberto, I., Gomes, E., & Da
has more accuracy in the prediction compared with traditional Silva, R. (2002). Optimization of xylanase production by Bacillus
RSM. circulans D1 in submerged fermentation using response surface
methodology. Process Biochemistry, 38(5), 727–731.
Box, G. E., & Behnken, D. W. (1960). Some new three level designs for
References the study of quantitative variables. Technometrics, 2(4), 455–475.
Box, G. E., & Wilson, K. (1951). On the experimental attainment of
Abdel-Fattah, Y. R. (2002). Optimization of thermostable lipase produc- optimum conditions. Journal of the Royal Statistical Society.
tion from a thermophilic Geobacillus sp. using Box-Behnken exper- Series B (Methodological), 13(1), 1–45.
imental design. Biotechnology Letters, 24(14), 1217–1222. Baranda, A. B., Etxebarria, N., Jimenez, R. M., & Alonso, R. M. (2005).
Abu-Izza, K. A., Garcia-Contreras, L., & Lu, D. R. (1996). Preparation Development of a liquid–liquid extraction procedure for five 1, 4-
and evaluation of sustained release AZT-loaded microspheres: opti- dihydropyridines calcium channel antagonists from human plasma
mization of the release characteristics using response surface meth- using experimental design. Talanta, 67(5), 933–941.
odology. Journal of Pharmaceutical Sciences, 85(2), 144–149. Bruns, R. E., Scarminio I. S., & de Barros Neto, B. (2006). Statistical
Açıkel, Ü., Erşan, M., & Açıkel, Y. S. (2010). Optimization of critical design-chemometrics (vol. 25): Elsevier
medium components using response surface methodology for lipase Burande, R. R., Kumbhar, B. K., Ghosh, P. K., & Jayas, D. S.
production by Rhizopus delemar. Food and Bioproducts (2008). Optimization of fluidized bed drying process of green
Processing, 88(1), 31–39. peas using response surface methodology. Drying Technology,
Acosta, O., Víquez, F., & Cubero, E. (2008). Optimisation of low calorie 26(7), 920–930.
mixed fruit jelly by response surface methodology. Food Quality Cao, H., Zhang, M., Mujumdar, A. S., Du, W. H., & Sun, J. C. (2006).
and Preference, 19(1), 79–85. Optimization of osmotic dehydration of kiwifruit. Drying
Ade-Omowaye, B., Rastogi, N., Angersbach, A., & Knorr, D. (2002). Technology, 24(1), 89–94.
Osmotic dehydration behavior of red paprika (Capsicum annuum Cheison, S. C., Wang, Z., & Xu, S.-Y. (2007). Use of response surface
L.). Journal of Food Science, 67(5), 1790–1796. methodology to optimise the hydrolysis of whey protein isolate in a
Agatonovic-Kustrin, S., Zecevic, M., Zivanovic, L., & Tucker, I. (1998). tangential flow filter membrane reactor. Journal of Food
Application of neural networks for response surface modeling in Engineering, 80(4), 1134–1145.
HPLC optimization. Analytica Chimica Acta, 364(1), 265–273. Chen, K.-N., Chen, M.-J., & Lin, C.-W. (2006). Optimal combination of
Ahn, J.-H., Kim, Y.-P., Lee, Y.-M., Seo, E.-M., Lee, K.-W., & Kim, H.-S. the encapsulating materials for probiotic microcapsules and its ex-
(2008). Optimization of microencapsulation of seed oil by response perimental verification (R1). Journal of Food Engineering, 76(3),
surface methodology. Food Chemistry, 107(1), 98–105. 313–320.
430 Food Bioprocess Technol (2017) 10:413–433

Chen, K. N., Chen, M. J., Liu, J. R., Lin, C. W., & Chiu, H. Y. (2005). Gan, C.-Y., Manaf, N. H. A., & Latiff, A. A. (2010). Optimization of
Optimization of incorporated prebiotics as coating materials for pro- alcohol insoluble polysaccharides (AIPS) extraction from the Parkia
biotic microencapsulation. Journal of Food Science, 70(5), M260– speciosa pod using response surface methodology (RSM).
M266. Carbohydrate Polymers, 79(4), 825–831.
Chen, Q., He, G., & Ali, M. A. (2002). Optimization of medium compo- Gangadharan, D., Sivaramakrishnan, S., Nampoothiri, K. M.,
sition for the production of elastase by Bacillus sp. EL31410 with Sukumaran, R. K., & Pandey, A. (2008). Response surface method-
response surface methodology. Enzyme and Microbial Technology, ology for the optimization of alpha amylase production by Bacillus
30(5), 667–672. amyloliquefaciens. Bioresource Technology, 99(11), 4597–4602.
Chen, X., Xu, F., Qin, W., Ma, L., & Zheng, Y. (2012). Optimization of Gao, L., & Mazza, G. (1996). Extraction of anthocyanin pigments from
enzymatic clarification of green asparagus juice using response sur- purple sunflower hulls. Journal of Food Science, 61(3), 600–603.
face methodology. Journal of Food Science, 77(6), C665–C670. Garrote, R. L., Silva, E. R., Bertone, R. A., & Roa, R. D. (2004).
Cladera-Olivera, F., Caron, G., & Brandelli, A. (2004). Bacteriocin pro- Predicting the end point of a blanching process. LWT-Food
duction by Bacillus licheniformis strain P40 in cheese whey using Science and Technology, 37(3), 309–315.
response surface methodology. Biochemical Engineering Journal, Ge, Y., Ni, Y., Yan, H., Chen, Y., & Cai, T. (2002). Optimization of the
21(1), 53–58. supercritical fluid extraction of natural vitamin E from wheat germ
Coman, G., & Bahrim, G. (2011). Optimization of xylanase produc- using response surface methodology. Journal of Food Science,
tion by Streptomyces sp. P12-137 using response surface meth- 67(1), 239–243.
odology and central composite design. Annals of Microbiology, Gfrerer, M., & Lankmayr, E. (2005). Screening, optimization and valida-
61(4), 773–779. tion of microwave-assisted extraction for the determination of per-
Cubas, C., Lobo, M. G., & González, M. (2008). Optimization of the sistent organochlorine pesticides. Analytica Chimica Acta, 533(2),
extraction of chlorophylls in green beans (Phaseolus vulgaris L.) 203–211.
by N, N-dimethylformamide using response surface methodology. Gharibzahedi, S. M. T., Mousavi, S. M., Hamedi, M., & Ghasemlou, M.
Journal of Food Composition and Analysis, 21(2), 125–133. (2012). Response surface modeling for optimization of formulation
Deepak, V., Kalishwaralal, K., Ramkumarpandian, S., Babu, S. V., variables and physical stability assessment of walnut oil-in-water
Senthilkumar, S., & Sangiliyandi, G. (2008). Optimization of media beverage emulsions. Food Hydrocolloids, 26(1), 293–301.
composition for Nattokinase production by Bacillus subtilis using Ghorbannezhad, P., Bay, A., Yolmeh, M., Yadollahi, R., & Moghadam, J.
response surface methodology. Bioresource Technology, 99(17), Y. (2016). Optimization of coagulation–flocculation process for me-
8170–8174. dium density fiberboard (MDF) wastewater through response sur-
Dutta, J. R., Dutta, P. K., & Banerjee, R. (2004). Optimization of culture face methodology. Desalination and Water Treatment, 1–16.
parameters for extracellular protease production from a newly iso- doi:10.1080/19443994.2016.1170636.
lated Pseudomonas sp. using response surface and artificial neural Gong, Y., Hou, Z., Gao, Y., Xue, Y., Liu, X., & Liu, G. (2012).
network models. Process Biochemistry, 39(12), 2193–2198. Optimization of extraction parameters of bioactive components
Erbay, Z., & Icier, F. (2009). Optimization of hot air drying of olive leaves from defatted marigold (Tagetes erecta L.) residue using re-
using response surface methodology. Journal of Food Engineering, s p o n s e s u r f a c e m e t h o d o l o g y. F o o d a n d B i o p ro d u c t s
91(4), 533–541. Processing, 90(1), 9–16.
Eren, İ., & Kaymak-Ertekin, F. (2007). Optimization of osmotic dehydra- Guo, W.-Q., Ren, N.-Q., Wang, X.-J., Xiang, W.-S., Ding, J., You, Y., &
tion of potato using response surface methodology. Journal of Food Liu, B.-F. (2009). Optimization of culture conditions for hydrogen
Engineering, 79(1), 344–352. production by Ethanoligenens harbinense B49 using response sur-
Esmaeili, M., Yolmeh, M., Shakerardakani, A., & Golivari, H. (2015). A face methodology. Bioresource Technology, 100(3), 1192–1196.
central composite design for the optimizing lipase and protease pro- Han, Q.-H., Yin, L.-J., Li, S.-J., Yang, B.-N., & Ma, J.-W. (2010).
duction from Bacillus subtilis PTCC 1720. Biocatalysis and Optimization of process parameters for microwave vacuum drying
Agricultural Biotechnology, 4(3), 349–354. of apple slices using response surface method. Drying Technology,
Fan, G., Han, Y., Gu, Z., & Chen, D. (2008). Optimizing conditions for 28(4), 523–532.
anthocyanins extraction from purple sweet potato using response He, Y.-Q., & Tan, T.-W. (2006). Use of response surface methodology to
surface methodology (RSM). LWT-Food Science and Technology, optimize culture medium for production of lipase with Candida sp.
41(1), 155–160. 99-125. Journal of Molecular Catalysis B: Enzymatic, 43(1), 9–14.
Felberg, I., Deliza, R., Farah, A., Calado, E., & Donangelo, C. (2010). Huang, Y.-B., Tsai, Y.-H., Lee, S.-H., Chang, J.-S., & Wu, P.-C. (2005).
Formulation of a soy–coffee beverage by response surface method- Optimization of pH-independent release of nicardipine hydrochlo-
ology and internal preference mapping. Journal of Sensory Studies, ride extended-release matrix tablets using response surface method-
25(s1), 226–242. ology. International Journal of Pharmaceutics, 289(1), 87–95.
Fernández, C., Dolores Alvarez, M., & Canet, W. (2006). The effect of Huynh, T. V., Caffin, N., Dykes, G. A., & Bhandari, B. (2008).
low-temperature blanching on the quality of fresh and frozen/thawed Optimization of the microencapsulation of lemon myrtle oil using
mashed potatoes. International journal of food science & technolo- response surface methodology. Drying Technology, 26(3), 357–368.
gy, 41(5), 577–595. Ismail, N., & Revathi, R. (2006). Studies on the effects of blanching time,
Francis, F., Sabu, A., Nampoothiri, K. M., Ramachandran, S., Ghosh, S., evaporation time, temperature and hydrocolloid on physical proper-
Szakacs, G., & Pandey, A. (2003). Use of response surface method- ties of chili (Capsicum annum var kulai) puree. LWT-Food Science
ology for optimizing process parameters for the production of α- and Technology, 39(1), 91–97.
amylase by Aspergillus oryzae. Biochemical Engineering Journal, Jackson, J. C., Bourne, M. C., & Barnard, J. (1996). Optimization of
15(2), 107–115. blanching for crispness of banana chips using response surface
Gallagher, E., O’Brien, C., Scannell, A., & Arendt, E. (2003). Use of methodology. Journal of Food Science, 61(1), 165–166.
response surface methodology to produce functional short dough Jafari, S. M., Assadpoor, E., He, Y., & Bhandari, B. (2008).
biscuits. Journal of Food Engineering, 56(2), 269–271. Encapsulation efficiency of food flavours and oils during spray dry-
Gan, C.-Y., & Latiff, A. A. (2011). Optimisation of the solvent extraction ing. Drying Technology, 26(7), 816–835.
of bioactive compounds from Parkia speciosa pod using response Juntachote, T., Berghofer, E., Bauer, F., & Siebenhandl, S. (2006). The
surface methodology. Food Chemistry, 124(3), 1277–1283. application of response surface methodology to the production of
Food Bioprocess Technol (2017) 10:413–433 431

phenolic extracts of lemon grass, galangal, holy basil and rosemary. Lin, Y.-P., Lee, T.-Y., Tsen, J.-H., & King, V. A.-E. (2007). Dehydration
International journal of food science & technology, 41(2), 121–133. of yam slices using FIR-assisted freeze drying. Journal of Food
Karazhiyan, H., Razavi, S. M., & Phillips, G. O. (2011). Extraction opti- Engineering, 79(4), 1295–1301.
mization of a hydrocolloid extract from cress seed (Lepidium Liu, J.-Z., Weng, L.-P., Zhang, Q.-L., Xu, H., & Ji, L.-N. (2003).
sati vum) u sing resp ons e surface met hod ol ogy. Food Optimization of glucose oxidase production by Aspergillus niger
Hydrocolloids, 25(5), 915–920. in a benchtop bioreactor using response surface methodology.
Kasankala, L. M., Xue, Y., Weilong, Y., Hong, S. D., & He, Q. (2007). World Journal of Microbiology and Biotechnology, 19(3), 317–323.
O p t i m i z a t i o n o f g e l a t i n e e x t r a c t i o n f r o m gr as s c a r p Lundstedt, T., Seifert, E., Abramo, L., Thelin, B., Nyström, Å., Pettersen,
(Catenopharyngodon idella) fish skin by response surface method- J., & Bergman, R. (1998). Experimental design and optimization.
ology. Bioresource Technology, 98(17), 3338–3343. Chemometrics and Intelligent Laboratory Systems, 42(1), 3–40.
Katapodis, P., Christakopoulou, V., Kekos, D., & Christakopoulos, P. Madamba, P. S. (2002). The response surface methodology: an applica-
(2007). Optimization of xylanase production by Chaetomium tion to optimize dehydration operations of selected agricultural
thermophilum in wheat straw using response surface methodology. crops. LWT-Food Science and Technology, 35(7), 584–592.
Biochemical Engineering Journal, 35(2), 136–141. Maran, J. P., & Manikandan, S. (2012). Response surface modeling and
Kaur, D., Wani, A. A., Oberoi, D., & Sogi, D. (2008). Effect of extraction optimization of process parameters for aqueous extraction of pig-
conditions on lycopene extractions from tomato processing waste ments from prickly pear (Opuntia ficus-indica) fruit. Dyes and
skin using response surface methodology. Food Chemistry, 108(2), Pigments, 95(3), 465–472.
711–718. Martinez, B., Rincón, F., Ibáñez, M., & BELLÁN, P. (2004). Improving
Kaur, S., Sarkar, B., Sharma, H., & Singh, C. (2009). Optimization of the nutritive value of homogenized infant foods using response sur-
enzymatic hydrolysis pretreatment conditions for enhanced juice face methodology. Journal of Food Science, 69(1), SNQ38–
recovery from guava fruit using response surface methodology. SNQ43.
Food and Bioprocess Technology, 2(1), 96–100. Masmoudi, M., Besbes, S., Chaabouni, M., Robert, C., Paquot, M.,
Kaushik, R., Saran, S., Isar, J., & Saxena, R. (2006). Statistical opti- Blecker, C., & Attia, H. (2008). Optimization of pectin extraction
mization of medium components and growth conditions by re- from lemon by-product with acidified date juice using response
sponse surface methodology to enhance lipase production by surface methodology. Carbohydrate Polymers, 74(2), 185–192.
Aspergillus carneus. Journal of Molecular Catalysis B: McCarthy, D., Gallagher, E., Gormley, T., Schober, T., & Arendt, E.
Enzymatic, 40(3), 121–126. (2005). Application of response surface methodology in the devel-
Kha, T. C., Nguyen, M. H., Roach, P. D., & Stathopoulos, C. E. (2014). opment of gluten-free bread. Cereal Chemistry, 82(5), 609–615.
Microencapsulation of gac oil by spray drying: optimization of wall Mestdagh, F., De Wilde, T., Fraselle, S., Govaert, Y., Ooghe, W.,
material concentration and oil load using response surface method- Degroodt, J.-M., & De Meulenaer, B. (2008). Optimization of the
ology. Drying Technology, 32(4), 385–397. blanching process to reduce acrylamide in fried potatoes. LWT-Food
Khazaei, K. M., Jafari, S., Ghorbani, M., Kakhki, A. H., & Sarfarazi, M. Science and Technology, 41(9), 1648–1654.
(2016). Optimization of anthocyanin extraction from saffron petals Mirmoghtadaie, L., Ensafi, A. A., Kadivar, M., & Norouzi, P. (2013).
with response surface methodology. Food Analytical Methods, 9(7), Highly selective electrochemical biosensor for the determination of
1993–2001. folic acid based on DNA modified-pencil graphite electrode using
Kim, M.-S., Kim, J.-S., You, Y.-H., Park, H. J., Lee, S., Park, J.-S., & response surface methodology. Materials Science and Engineering:
Hwang, S.-J. (2007). Development and optimization of a novel oral C, 33(3), 1753–1758.
controlled delivery system for tamsulosin hydrochloride using re- Montgomery, D. C. (2008). Design and analysis of experiments: John
sponse surface methodology. International Journal of Wiley & Sons.
Pharmaceutics, 341(1), 97–104. Moulai Mostefa, N., Hadj Sadok, A., Sabri, N., & Hadji, A. (2006).
Ko, J., Park, H. J., Park, Y., Hwang, S., & Park, J. (2003). Chitosan Determination of optimal cream formulation from long-term stabil-
microparticle preparation for controlled drug release by response ity investigation using a surface response modelling. International
surface methodology. Journal of Microencapsulation, 20(6), Journal of Cosmetic Science, 28(3), 211–218.
791–797. Moyo, S., Gashe, B., Collison, E., & Mpuchane, S. (2003). Optimising
Kumar, D., Prasad, S., & Murthy, G. S. (2014). Optimization of growth conditions for the pectinolytic activity of Kluyveromyces
microwave-assisted hot air drying conditions of okra using response wickerhamii by using response surface methodology. International
surface methodology. Journal of Food Science and Technology, Journal of Food Microbiology, 85(1), 87–100.
51(2), 221–232. Muralidhar, R., Chirumamila, R., Marchant, R., & Nigam, P. (2001). A
Kunamneni, A., & Singh, S. (2005). Response surface optimization of response surface approach for the comparison of lipase production
enzymatic hydrolysis of maize starch for higher glucose production. by Candida cylindracea using two different carbon sources.
Biochemical Engineering Journal, 27(2), 179–190. Biochemical Engineering Journal, 9(1), 17–23.
Kurozawa, L., Park, K., & Hubinger, M. (2008). Optimization of the Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. M. (2016).
enzymatic hydrolysis of chicken meat using response surface meth- Response surface methodology: process and product optimization
odology. Journal of Food Science, 73(5), C405–C412. using designed experiments: John Wiley & Sons.
Lee, W., Yusof, S., Hamid, N. S. A., & Baharin, B. S. (2006). Optimizing Mylonaki, S., Kiassos, E., Makris, D. P., & Kefalas, P. (2008).
conditions for enzymatic clarification of banana juice using response Optimisation of the extraction of olive (Olea europaea) leaf phe-
surface methodology (RSM). Journal of Food Engineering, 73(1), nolics using water/ethanol-based solvent systems and response
55–63. surface methodology. Analytical and Bioanalytical Chemistry,
Li, X., Liu, Z., & Chi, Z. (2008). Production of phytase by a marine yeast 392(5), 977–985.
Kodamaea ohmeri BG3 in an oats medium: optimization by re- Neifar, M., ELLOUZE-GHORBEL, R., Kamoun, A., Baklouti, S.,
sponse surface methodology. Bioresource Technology, 99(14), Mokni, A., Jaouani, A., & ELLOUZE-CHAABOUNI, S. (2011).
6386–6390. Effective clarification of pomegranate juice using laccase treatment
Liew, S. L., Ariff, A., Raha, A., & Ho, Y. (2005). Optimization of medium optimized by response surface methodology followed by ultrafiltra-
composition for the production of a probiotic microorganism, tion. Journal of Food Process Engineering, 34(4), 1199–1219.
Lactobacillus rhamnosus, using response surface methodology. Nikerel, İ. E., Öner, E., Kirdar, B., & Yildirim, R. (2006). Optimization of
International Journal of Food Microbiology, 102(2), 137–142. medium composition for biomass production of recombinant
432 Food Bioprocess Technol (2017) 10:413–433

Escherichia coli cells using response surface methodology. Saccani, G., Tanzi, E., Mallozzi, S., & Cavalli, S. (2005). Determination
Biochemical Engineering Journal, 32(1), 1–6. of niacin in fresh and dry cured pork products by ion chromatogra-
Novotná, K., Havliš, J., & Havel, J. (2005). Optimisation of high perfor- phy: experimental design approach for the optimisation of nicotinic
mance liquid chromatography separation of neuroprotective pep- acid separation. Food Chemistry, 92(2), 373–379.
tides: fractional experimental designs combined with artificial neural Salimi, A., Maghsoudlou, Y., Jafari, S. M., Mahoonak, A. S.,
networks. Journal of Chromatography A, 1096(1), 50–57. Kashaninejad, M., & Ziaiifar, A. M. (2015). Preparation of lycopene
Nur’Aliaa, A., Siti Mazlina, M., Taip, F. S., & Liew Abdullah, A. (2010). emulsions by whey protein concentrate and maltodextrin and opti-
Response surface optimization for clarification of white pitaya juice mization by response surface methodology. Journal of Dispersion
using a commercial enzyme. Journal of Food Process Engineering, Science and Technology, 36(2), 274–283.
33(2), 333–347. Sanchez, H., Osella, C., & De La Torre, M. (2004). Use of response
Ozdemir, M., Ozen, B. F., Dock, L. L., & Floros, J. D. (2008). surface methodology to optimize gluten-free bread fortified with
Optimization of osmotic dehydration of diced green peppers by soy flour and dry milk. Food Science and Technology
response surface methodology. LWT-Food Science and International, 10(1), 5–9.
Technology, 41(10), 2044–2050. Sarfarazi, M., Jafari, S. M., & Rajabzadeh, G. (2015). Extraction optimi-
Pappa, I., Bloukas, J., & Arvanitoyannis, I. (2000). Optimization of salt, zation of saffron nutraceuticals through response surface methodol-
olive oil and pectin level for low-fat frankfurters produced by re- ogy. Food Analytical Methods, 8(9), 2273–2285.
placing pork backfat with olive oil. Meat Science, 56(1), 81–88. Schwabe, C., & Büllesbach, E. (2013). Relaxin and the fine structure of
Peričin, D., Radulović-Popović, L., Vaštag, Ž., Mađarev-Popović, S., & proteins: Springer Science & Business Media.
Trivić, S. (2009). Enzymatic hydrolysis of protein isolate from hull- Servais, A. C., Fillet, M., Chiap, P., Abushoffa, A. M., Hubert, P., &
less pumpkin oil cake: application of response surface methodology. Crommen, J. (2002). Optimization of the separation of β-blockers
Food Chemistry, 115(2), 753–757. by ion-pair capillary electrophoresis in non-aqueous media using
Pompeu, D., Silva, E., & Rogez, H. (2009). Optimisation of the solvent univariate and multivariate approaches. Journal of Separation
extraction of phenolic antioxidants from fruits of Euterpe oleracea Science, 25(15–17), 1087–1095.
using response surface methodology. Bioresource Technology, Shahraki, M. H., Jafari, S. M., Mashkour, M., & Esmaeilzadeh, E. (2013).
100(23), 6076–6082. Optimization of closed-cycle fluidized bed drying of sesame seeds
using response surface methodology and genetic algorithms.
Puri, S., Beg, Q. K., & Gupta, R. (2002). Optimization of alkaline prote-
International Journal of Food Engineering, 10(1), 167–181.
ase production from Bacillus sp. by response surface methodology.
Sin, H., Yusof, S., Hamid, N. S. A., & Rahman, R. A. (2006a).
Current Microbiology, 44(4), 286–290.
Optimization of hot water extraction for sapodilla juice using re-
Quintero-Ramos, A., Bourne, M., Barnard, J., & ANZALDÚA-
sponse surface methodology. Journal of Food Engineering, 74(3),
MORALES, A. (1998). Optimization of low temperature blanching
352–358.
of frozen jalapeño pepper (Capsicum annuum) using response sur-
Sin, H. N., Yusof, S., Hamid, N. S. A., & Rahman, R. A. (2006b).
face methodology. Journal of Food Science, 63(3), 519–522.
Optimization of enzymatic clarification of sapodilla juice using re-
Rai, P., Majumdar, G., Dasgupta, S., & De, S. (2004). Optimizing
sponse surface methodology. Journal of Food Engineering, 73(4),
pectinase usage in pretreatment of mosambi juice for clarification
313–319.
by response surface methodology. Journal of Food Engineering,
Singh, B., Chakkal, S. K., & Ahuja, N. (2006). Formulation and
64(3), 397–403.
optimization of controlled release mucoadhesive tablets of aten-
Ratnam, B., Rao, M. N., Rao, M. D., Rao, S. S., & Ayyanna, C. (2003). olol using response surface methodology. AAPS PharmSciTech,
Optimization of fermentation conditions for the production of etha- 7(1), E19–E28.
nol from sago starch using response surface methodology. World Singh, B., Panesar, P. S., Gupta, A., & Kennedy, J. F. (2007).
Journal of Microbiology and Biotechnology, 19(5), 523–526. Optimisation of osmotic dehydration of carrot cubes in sucrose-
Reyes-Moreno, C., Parra-Inzunza, M. A., Milán-Carrillo, J., & Zazueta- salt solutions using response surface methodology. European Food
Niebla, J. A. (2002). A response surface methodology approach to Research and Technology, 225(2), 157–165.
optimise pretreatments to prevent enzymatic browning in potato Singh, S., Raina, C. S., Bawa, A. S., & Saxena, D. C. (2004). Sweet
(Solanum tuberosum L) cubes. Journal of the Science of Food and potato-based pasta product: optimization of ingredient levels using
Agriculture, 82(1), 69–79. response surface methodology. International journal of food science
Rezzoug, S.-A., Boutekedjiret, C., & Allaf, K. (2005). Optimization of & technology, 39(2), 191–200.
operating conditions of rosemary essential oil extraction by a fast Sinha, K., Chowdhury, S., Saha, P. D., & Datta, S. (2013). Modeling of
controlled pressure drop process using response surface methodol- microwave-assisted extraction of natural dye from seeds of Bixa
ogy. Journal of Food Engineering, 71(1), 9–17. orellana (Annatto) using response surface methodology (RSM)
Rodrigues, L., Teixeira, J., Oliveira, R., & Van Der Mei, H. C. (2006). and artificial neural network (ANN). Industrial Crops and
Response surface optimization of the medium components for the Products, 41, 165–171.
production of biosurfactants by probiotic bacteria. Process Sinha, K., Saha, P. D., & Datta, S. (2012). Extraction of natural dye from
Biochemistry, 41(1), 1–10. petals of flame of forest (Butea monosperma) flower: process opti-
Rodrigues, S., Pinto, G. A., & Fernandes, F. A. (2008). Optimization of mization using response surface methodology (RSM). Dyes and
ultrasound extraction of phenolic compounds from coconut (Cocos Pigments, 94(2), 212–216.
nucifera) shell powder by response surface methodology. Souza, A. S., dos Santos, W. N., & Ferreira, S. L. (2005). Application of
Ultrasonics Sonochemistry, 15(1), 95–100. Box–Behnken design in the optimisation of an on-line pre-concen-
Rodríguez-Nogales, J. M., Ortega, N., Perez-Mateos, M., & Busto, M. D. tration system using knotted reactor for cadmium determination by
(2007). Experimental design and response surface modeling applied flame atomic absorption spectrometry. Spectrochimica Acta Part B:
for the optimisation of pectin hydrolysis by enzymes from A. niger Atomic Spectroscopy, 60(5), 737–742.
CECT 2088. Food Chemistry, 101(2), 634–642. Spanilá, M., Pazourek, J., Farková, M., & Havel, J. (2005). Optimization
Rosa, P. A., Azevedo, A. M., & Aires-Barros, M. R. (2007). Application of solid-phase extraction using artificial neural networks in combi-
of central composite design to the optimisation of aqueous two- nation with experimental design for determination of resveratrol by
phase extraction of human antibodies. Journal of Chromatography capillary zone electrophoresis in wines. Journal of Chromatography
A, 1141(1), 50–60. A, 1084(1), 180–185.
Food Bioprocess Technol (2017) 10:413–433 433

Stamatopoulos, K., Katsoyannos, E., Chatzilazarou, A., & Konteles, Wang, L., Yang, B., Du, X., Yang, Y., & Liu, J. (2008). Optimization of
S. J. (2012). Improvement of oleuropein extractability by conditions for extraction of acid-soluble collagen from grass carp
optimising steam blanching process as pre-treatment of olive leaf (Ctenopharyngodon idella) by response surface methodology.
extraction via response surface methodology. Food Chemistry, Innovative Food Science & Emerging Technologies, 9(4), 604–607.
133(2), 344–351. Wijngaard, H. H., & Brunton, N. (2010). The optimisation of solid–liquid
Suh, M.-H., Yoo, S.-H., & Lee, H. G. (2007). Antioxidative activity and extraction of antioxidants from apple pomace by response surface
structural stability of microencapsulated γ-oryzanol in heat-treated methodology. Journal of Food Engineering, 96(1), 134–140.
lards. Food Chemistry, 100(3), 1065–1070. Xie, J.-H., Shen, M.-Y., Xie, M.-Y., Nie, S.-P., Chen, Y., Li, C., & Wang,
Sunitha, K., Lee, J.-K., & Oh, T.-K. (1999). Optimization of medium Y.-X. (2012). Ultrasonic-assisted extraction, antimicrobial and anti-
components for phytase production by E. coli using response sur- oxidant activities of Cyclocarya paliurus (Batal.) Iljinskaja polysac-
face methodology. Bioprocess Engineering, 21(6), 477–481. charides. Carbohydrate Polymers, 89(1), 177–184.
Suresh Babu, C. V., Chung, B. C., & Yoo, Y. S. (2004). Experimental Xie, J. H., Xie, M. Y., Shen, M. Y., Nie, S. P., Li, C., & Wang, Y. X.
design to investigate factors affecting capillary zone electrophoresis. (2010). Optimisation of microwave-assisted extraction of polysac-
Analytical Letters, 37(12), 2485–2499. charides from Cyclocarya paliurus (Batal.) Iljinskaja using response
Tabaraki, R., & Nateghi, A. (2011). Optimization of ultrasonic-assisted surface methodology. Journal of the Science of Food and
extraction of natural antioxidants from rice bran using response sur- Agriculture, 90(8), 1353–1360.
face methodology. Ultrasonics Sonochemistry, 18(6), 1279–1286. Xiong, C., Jinhua, W., & Dongsheng, L. (2007). Optimization of solid-
Tang, X.-J., He, G.-Q., Chen, Q.-H., Zhang, X.-Y., & Ali, M. A. (2004). state medium for the production of inulinase by Kluyveromyces
Medium optimization for the production of thermal stable β- S120 using response surface methodology. Biochemical
glucanase by Bacillus subtilis ZJF-1A5 using response surface Engineering Journal, 34(2), 179–184.
methodology. Bioresource Technology, 93(2), 175–181.
Yang, L., Jiang, J. G., Li, W. F., Chen, J., Wang, D. Y., & Zhu, L. (2009).
Tanyildizi, M. S., Özer, D., & Elibol, M. (2005). Optimization of α-
Optimum extraction process of polyphenols from the bark of
amylase production by Bacillus sp. using response surface method-
Phyllanthus emblica L. based on the response surface methodology.
ology. Process Biochemistry, 40(7), 2291–2296.
Journal of Separation Science, 32(9), 1437–1444.
Thakur, S., & Saxena, D. (2000). Formulation of extruded snack food
(gum based cereal–pulse blend): optimization of ingredients levels Yolmeh, M., Najafi, M. B. H., & Farhoosh, R. (2014). Optimisation of
using response surface methodology. LWT-Food Science and ultrasound-assisted extraction of natural pigment from annatto seeds
Technology, 33(5), 354–361. by response surface methodology (RSM). Food Chemistry, 155,
Tonon, R. V., Grosso, C. R., & Hubinger, M. D. (2011). Influence of 319–324.
emulsion composition and inlet air temperature on the microencap- Yolmeh, M., & Najafzadeh, M. (2014). Optimisation and modelling
sulation of flaxseed oil by spray drying. Food Research green bean’s ultrasound blanching. International journal of food
International, 44(1), 282–289. science & technology, 49(12), 2678–2684.
Tran, A. T., Hyne, R. V., Pablo, F., Day, W. R., & Doble, P. (2007). Yoshida, T., Tsubaki, S., Teramoto, Y., & Azuma, J. I. (2010).
Optimisation of the separation of herbicides by linear gradient high Optimization of microwave-assisted extraction of carbohydrates
performance liquid chromatography utilising artificial neural net- from industrial waste of corn starch production using response sur-
works. Talanta, 71(3), 1268–1275. face methodology. Bioresource Technology, 101(20), 7820–7826.
Vohra, A., & Satyanarayana, T. (2002). Statistical optimization of the medi- Zhang, Q., Qu, Y., Zhou, J., Zhang, X., Zhou, H., Ma, Q., & Li, X. (2011).
um components by response surface methodology to enhance phytase Optimization of 2,3-dihydroxybiphenyl 1,2-dioxygenase expression
production by Pichia anomala. Process Biochemistry, 37(9), 999–1004. and its application for biosensor. Bioresource Technology, 102,
Wang, H., Liu, Y., Wei, S., & Yan, Z. (2012). Application of response 10553–10560.
surface methodology to optimise supercritical carbon dioxide ex- Zhong, K., & Wang, Q. (2010). Optimization of ultrasonic extraction of
traction of essential oil from Cyperus rotundus Linn. Food polysaccharides from dried longan pulp using response surface
Chemistry, 132(1), 582–587. methodology. Carbohydrate Polymers, 80(1), 19–25.

You might also like