International Journal of Food Science and Technology 2004, 39, 191–200 191

Sweet potato-based pasta product: optimization of

ingredient levels using response surface methodology

Sukhcharn Singh,1 Charanjit Singh Raina,1 Amrinder Singh Bawa2 & Dharmesh Chandra Saxena1*
1 Department of Food Technology, Sant Longowal Institute of Engineering and Technology, Longowal 148 106, India
2 Defence Food Research Laboratory, Mysore, India
(Received 29 September 2001; Accepted in revised form 25 June 2003)

Summary Sweet potato flour was used for the development of a pasta product. The system known as
response surface methodology was used to analyse the effect of sweet potato flour,
soyflour, water, Arabic gum and carboxy methyl cellulose (CMC) on quality responses
(sensory, solids loss and hardness) of the pasta product. A rotatable central-composite
design was used to develop models for the responses. Responses were affected most by
changes in soyflour and gum levels and to a lesser extent by sweet potato flour and water
levels. Individual contour plots of the different responses were superimposed and regions
meeting the maximum sensory score (33.8), minimum solids loss (16.6%) and maximum
texture hardness (5616 g) were identified at 674 g kg)1 sweet potato flour, 195 g kg)1
water, 110 g kg)1 soyflour, 10.6 g kg)1 Arabic gum and 10.1 g kg)1 CMC levels.
Keywords Arabic gum, carboxy methyl cellulose, cooking loss, sensory analysis, soyflour, texture.

providing at least 90% of the requirement

Introduction
(Recommended Dietary Allowance) for all items
Sweet potato, botanically known as Ipomoea except protein and niacin (Bouwkamp, 1985). In
batatas (L.), is a variety of orange sweet potatoes. combination with legumes and other protein-rich
Over 95% of the global sweet potato crop is food, sweet potato could be used to combat
produced in developing countries, where it is the malnutrition. The total consumption of sweet
fifth most important food crop in terms of fresh potato is very low at present, but it could be
weight. Asia is the world’s largest sweet potato- increased substantially if appropriate and inex-
producing region, with 125 million tons of annual pensive technologies were introduced to process it.
production. India is the fourth largest producer Hamed et al. (1973) have found the sweet
and produces about 1.3 million tonnes of sweet potato flour can be used as a substitute for wheat
potato compared with the world’s production of flour in bread-making. Flours from peeled and
133 million tonnes per year. Nearly half of the unpeeled sweet potatoes were found to be equiv-
sweet potatoes produced in Asia are used for alent in yield and colour, but flour from the latter
animal feed, with the remainder primarily used for was higher in ash and crude fibre. Several workers
human consumption, either as fresh or processed have described the preparation of the flour under
products (Centre Proposals and TAC Recommen- different conditions. Lizada & Guzman (1982)
dations, 2000). Sweet potato has a large potential found that the drying of sweet potato slices in a
to be used as a food in developing nations with cabinet drier at a temperature of 60–70
C was
limited resources because of its short maturity able to yield flour with a moisture content of
time and ability to grow under diverse climatic 6.6%. When sweet potato meal was partially
condition and on less fertile soil. Sweet potato replaced (from 10 to 90%) with soya meal it was
roots are an excellent source of all nutrients, found that there was decreased water absorption
capacity and bulk density (Iwe & Onuh, 1992).
*Correspondent: Fax: 91 1672 280057; El-sahy & Siliha (1988) supplemented wheat flour
e-mail: [email protected] with sulphited sweet potato flour (from 5 to 20%)

 2004 Blackwell Publishing Ltd

192 Optimization of pasta made from sweet potato S. Singh et al.

and found that supplementation improved the enhance the nutritional protein quality of pasta.
sensory properties of Egyptian bread. Chiappe The production of sweet potato per hectare is also
et al. (1984) found that substituting 10% of the higher than that of the cereals used for making
wheat with sweet potato flour improved the pasta.
quality of bread. In India, the size of the semi-processed/ready-
Defatted soyflour and grits are the most to-eat food segment is over Rs. 4000 crores (US
economical sources of soya protein available and $1.1 billion), with over 60 000 bakeries, 20 000
the nutritional value of these products is some- traditional food units and several pasta food units
what superior to that of concentrated (refined) (the annual production of macaroni/noodles is
protein products made from soybean. Limroong- about 16 500 tonnes). The annual production of
reungrat & Huang (2002) observed that pasta bread, biscuits and other bakery products in India
made from 100% alkaline-treated sweet potato is estimated to be 3.70 million tonnes (Ministry of
flour had the lowest cooking loss (9.9%) and Food Processing Industries, 2003).
the highest firmness (1.8 N). Different hydrocol- However, the standardization of the process and
loids (sodium alginate, K-carrageenan, xanthan formulation of pasta using sweet potato flour has
gum and hydroxypropyl methylcellulose) have not been reported in the literature. Therefore, in
improved the rheological properties of the wheat the present study, investigations were undertaken
flour dough and the final quality of bread (Rosell to utilize the easy availability of sweet potato to
et al., 2001). develop a pasta product in combination with
Response surface methodology (RSM) has been soyflour and hydrocolloids and to use RSM to
shown to be an effective tool for optimizing a optimize the formulation of raw ingredients. The
process as highlighted by different authors. Basic- production of a high quality soya sweet potato
ally RSM relates product properties by using blend-based pasta product was dependant on
regression equations that describe interrelations different responses such as sensory quality, solids
between input parameters and product properties loss and textural property being measured and
(Colonna et al., 1984). Jauregui et al. (2000) used optimized.
RSM to optimize the conditions for an extruder
producing an amaranth-based snack food. Thakur
Materials
& Saxena (2000) used RSM to analyse the effect of
corn flour, green gram flour, xanthan, guar gum, Sweet potatoes were purchased from a local
Arabic gum and carboxy methyl cellulose (CMC) market and its slices were lye peeled, dried at
on the sensory and objective (expansion ratio) 65
C to 12% moisture content, ground and then
attributes of an extruded snack food and found passed through an eighty-mesh sieve to make
that the responses were affected most by changes flour. The flour contained 1.56 g kg)1 ash and
in corn flour, green gram flour and guar gum levels 2.3 g kg)1 protein. The flour was kept at 4
C
and to a lesser extent by xanthan, Arabic gum and until required.
CMC levels. Vatsala et al. (2001) optimized the Defatted soyflour was kindly donated by Mod-
ingredients and process conditions for the prepar- ern Food Industries, Chandigarh, India. The
ation of puri using RSM. Some good examples of soyflour contained 8% moisture, 1.2 g kg)1 ash
the appropriate applications of this technique in and 57.5 g kg)1 protein. Arabic gum was procured
textured products are the optimization of complex from the local market. Laboratory grade CMC
products or properties or many process variables was purchased from Brightways Agencies,
(Onwulata et al., 2000; Sebio & Chang, 2000; Ambala, India.
Rosell et al., 2001). A pasta-making machine (model: Dolly; La
The nutritional value of pasta, as shown by its Monforrina, Asti, Italy) was used in the present
bromatological composition, is not very high, as it study, in which relatively dry dough is forced into
is rich in starch, whereas its protein concentration the holes in the die under a pressure of approxi-
and quality is significantly lower (approximately mately 6895 kPa. A standard weight of raw
seven to eight times). Therefore, the incorporation material (256 g) was used for each experimental
of soyflour in the pasta preparation would run.

International Journal of Food Science and Technology 2004, 39, 191–200  2004 Blackwell Publishing Ltd
 Optimization of pasta made from sweet potato S. Singh et al. 193

pieces were dried at 60
C for 30 min to a final

Methods
moisture content of c. 12% and packed in
commercially available low-density polythene
Pasta preparation
pouches.
The flow sheet of the pasta-making process is
shown in Fig. 1. The production process for the
Estimation of the responses
pasta products consisted of adding water to the
raw ingredients to make a mixture of about 31% The samples were tested to establish the maximum
moisture content, mixing and kneading the mass sensory score, texture (hardness) and minimum
for 15 min to produce a stiff, plastic, homo- solids loss after cooking (in the boiled water). The
geneous dough and then applying pressure so as to values are averages of triplicate determinations.
extrude the dough through a series of die orifices.
The extruded products (Macaroni, Rigatoni
Cooking time
Form – a tubular shape with a ribbed surface, of
0.92 mm thickness, 5.3 mm external diameter Product (5 g dry wt) was added to 200 mL of
and 32.0 mm length) were cut into pieces of rapidly boiling water. The product was deemed
uniform size by means of a knife moving over the cooked when the white core in the pasta, when
outer die surface. After extrusion, the shaped pressed between two glass slides, was no longer
visible. The time required to reach this point for
each sample, denoted as cooking time, was 5 min.
SWEET POTATO
Solids loss
GRADING and WASHING
Ten grams of sample was weighed in a beaker and
200 mL of boiling water was added to it. The
LYE PEELING (8% NaOH, for 4 min)
beaker was placed in a boiling water bath for a
minimum cooking time of 5 min. The cooking
SLICING (1.3 – 1.5 cm thick)
water was drained into a 250-mL volumetric flask
and the volume was made up to 250 mL. Ten
TREATMENT (2% Potassium metabisulphite, millilitres of the cooking water was evaporated in
for 30 min)
an air oven at 100
C to determine the weight (A)
DRYING (65˚C) of the residue.
Solids loss ð%Þ ¼ AB  100=CD;
GRINDING (Sieve size 80 mesh) where A ¼ weight of residue, B ¼ volume of
cooking water (250 mL), C ¼ initial sample
INGREDIENTS MIXING (As per RSM design) weight (10 g), and D ¼ volume of sample taken
for residue estimation (10 mL).
KNEADING (for 15 min)

Sensory evaluation
EXTRUSION
The sensory analysis of the cooked product was
PASTA PRODUCT done by a panel of five people. The samples were
evaluated for colour, appearance (deformation),
PARTIAL DRYING (60˚C for 30 min)
taste, aroma and mouth-feel. Each of the five
attributes was evaluated using a nine-point scoring
scale, with a liked very much product, retaining its
PACKING and STORAGE
original shape and with a smooth, chewable, non-
Figure 1 Flow sheet for the pasta preparation from sweet sticky mouth-feel, scoring 9, and a disliked very
potato flour. much, extensively deformed product, scoring 1.

 2004 Blackwell Publishing Ltd International Journal of Food Science and Technology 2004, 39, 191–200
194 Optimization of pasta made from sweet potato S. Singh et al.

Selecting values for each of the independent ratios

Determination of texture (hardness)
uniquely determined the amounts of the five
The texture (hardness) of the uncooked pasta ingredients. The constraint of transforming to
was measured using a texture analyser (model ratios complicates interpretation of results but
TA-XT2; Stable Micro Systems, Haslemere, UK) permits a mathematical approximation to the
under the following conditions: mode, force in pasta quality–composition relationship.
compression; load cell, 25 kg; pre-test speed, A centre point for the design was selected with
2 mm s)1; post-test speed, 10 mm s)1; distance, ingredients at levels expected to yield, at least,
5 mm; trigger, 5 g. satisfactory experimental results. With the centre
composition selected, the normal xi ratios were
calculated by using the normal weight composi-
Experimental design
tion of the formulation given in Table 1. The
The study was based on the hypothesis that design depended upon the symmetrical selection of
sensory score, solids loss and texture of the variation increments about the centre composi-
product were functionally related to specific com- tion. These levels of variation were chosen to be
position, and attempts were made to fit multiple within the range of reasonable formulation, and
regression equation describing responses to qual- the increments were carefully selected, as inter-
ity composition. pretation of the result was valid only within the
Table 1 lists ingredients in the descending order experimental limits. The levels selected were also
of assumed importance as a quality variable. The based on the conclusions of previous studies.
composition of a pasta had the form: A (sweet The increments of variation for each variable
potato flour) + B (water) + C (soyflour) + D spaced around the centre point ratios, along with
(Arabic gum) + E (CMC) ¼ 100%. This equa- the equations relating the actual and coded ratios,
tion implies mathematical linear dependence of are presented in Table 2. By substituting these
the variables if the amounts of ingredients are equations, compositions were coded for solution
used directly as variables, as, from the equation, of the multiple regression equations.
the quantity of any ingredient is uniquely deter-
mined by the amounts of the other four. To
Practical considerations
function in a multiple factor analysis, these
ingredients may be transformed to ratios, which Before this type of experiment could be done, the
can be varied independently. For this experiment, coded Xi ratios for each treatment as per the
the following ingredient ratios were selected as the experimental design were translated into working
xi variable. quantities of ingredients. The pasta compositions
were obtained by systematic algebraic solutions
x1 ¼ A=ðB þ C þ D þ EÞ ¼ 1:98;
for A, B, C, D and E in terms of actual xi ratios
x2 ¼ B=ðC þ D þ EÞ ¼ 1:39; and a unit quantity of product. Equations derived
x3 ¼ C=ðD þ EÞ ¼ 5:00; for the general case follow:
x4 ¼ D=E ¼ 1:00:

Table 2 Experimental increments, values of coded levels and

Table 1 Formula composition at the design centre point
equations relating actual xi and coded Xi ratios
Weight
Xi coded levels
Total mixture
Ingredients basis (%) g g kg)1
xi Variable ±Increment )2 )1 0 +1 +2
A. Sweet potato flour 66.44 170.09 664.43
x1 0.10 1.78 1.88 1.98 2.08 2.18
B. Water 19.52 49.96 195.16
x2 0.10 1.19 1.29 1.39 1.49 1.59
C. Soyflour 11.70 29.95 117.01
x3 0.50 4.00 4.50 5.00 5.50 6.00
D. Arabic gum 1.17 3.00 11.70
x4 0.20 0.60 0.80 1.00 1.20 1.40
E. CMC 1.17 3.00 11.70

Total 100.00 256.00 1000.0 where X1 ¼ (x1 – 1.98)/0.1; X2 ¼ (x2 – 1.39)/0.1; X3 ¼ (x3 –
5.00)/0.5; X4 ¼ (x4 – 1.00)/0.2.

International Journal of Food Science and Technology 2004, 39, 191–200  2004 Blackwell Publishing Ltd
 Optimization of pasta made from sweet potato S. Singh et al. 195

x1 x2 ð1  AÞ existed for each response variable, Yk, in terms

A¼ ; B¼ ;
1 þ x1 1 þ x2 of m independent processing factors Xi
x3 ð1  A  BÞ x4 ð1  A  B  CÞ (i ¼ 1,2,…,m):
C¼ ; D¼ ;
1 þ x3 1 þ x4 Yk ¼ fk ðX1 ; X2 ; . . . ; Xm Þ:
D
E¼ In this case, n ¼ 3 and m ¼ 4. The function was
x4
assumed to be approximated by a second-degree
The resulting weights for each ingredient in polynomial equation:
different treatments are given in Table 3. A central X4 X
4 X
4

composite rotatable design was adopted (Cocha- Yk ¼ bk0 þ bki Xi þ bkii X2i þ bkij Xi Xj
i¼1 i¼1 i6¼j¼1
ran & Cox, 1957). In this design, experiments were
randomized in order to minimize the effects of ð1Þ
unexplained variability in the observed responses where bk0 is the value of the fitted response at the
due to extraneous factors. For analysis of the centre point of the design, i.e. point (0,0,0,0), and
experimental design by the RSM, it was assumed bki, bkii and bkij are the linear, quadratic and cross-
that n mathematical functions, fk (k ¼ 1, 2,…,n), product regression terms, respectively.

Table 3 Central composite design

arrangement and responses Treatment
coded level Composition (g) Responses

Expt. no. X1 X2 X3 X4 A B C D E Y1 Y2 Y3

1 )1 )1 )1 )1 167.11 50.07 31.76 3.14 3.92 36.12 17.98 6718.54

2 )1 )1 )1 +1 167.11 50.07 31.76 3.85 3.21 35.50 19.75 3789.36
3 )1 )1 +1 )1 167.11 50.07 32.84 2.65 3.32 35.70 19.92 3105.17
4 )1 )1 +1 +1 167.11 50.07 32.84 3.26 2.71 36.25 22.63 4481.48
5 )1 +1 )1 )1 167.11 53.19 29.21 2.88 3.61 36.00 25.82 6628.37
6 )1 +1 )1 +1 167.11 53.19 29.21 3.54 2.95 34.30 20.16 6027.92
7 )1 +1 +1 )1 167.11 53.19 30.21 2.44 3.05 31.60 20.18 7284.33
8 )1 +1 +1 +1 167.11 53.19 30.21 4.65 0.84 31.00 17.56 6665.40
9 +1 )1 )1 )1 172.88 46.82 29.70 2.93 3.67 35.60 17.84 5952.38
10 +1 )1 )1 +1 172.88 46.82 29.70 3.60 3.00 34.66 18.19 448.92
11 +1 )1 +1 )1 172.88 46.82 30.71 2.48 3.10 35.00 28.62 8188.92
12 +1 )1 +1 +1 172.88 46.82 30.71 3.05 2.54 36.12 29.92 5665.59
13 +1 +1 )1 )1 172.88 49.74 27.31 2.70 3.37 35.28 18.20 333.27
14 +1 +1 )1 +1 172.88 49.74 27.31 3.31 2.76 35.00 13.57 1338.47
15 +1 +1 +1 )1 172.88 49.74 28.24 2.28 2.85 33.85 22.51 9315.51
16 +1 +1 +1 +1 172.88 49.74 28.24 2.80 2.33 33.24 17.43 6399.25
17 )2 0 0 0 163.91 53.56 32.11 3.21 3.21 36.21 20.06 4468.75
18 +2 0 0 0 175.50 46.82 28.07 2.81 2.81 35.31 19.70 3500.00
19 0 )2 0 0 170.09 46.68 32.69 3.27 3.27 36.50 16.90 4218.75
20 0 +2 0 0 170.09 52.74 27.64 2.76 2.76 34.01 16.49 2428.57
21 0 0 )2 0 170.09 49.96 28.76 3.59 3.59 35.50 19.90 4332.51
22 0 0 +2 0 170.09 49.96 30.81 2.57 2.57 32.51 23.59 8903.07
23 0 0 0 )2 170.09 49.96 29.95 2.25 3.74 34.68 19.90 4356.87
24 0 0 0 +2 170.09 49.96 29.95 3.49 2.50 32.50 19.52 3984.21
25 0 0 0 0 170.09 49.96 29.95 3.00 3.00 34.75 16.83 3655.14
26 0 0 0 0 170.09 49.96 29.95 3.00 3.00 35.25 16.53 4421.25
27 0 0 0 0 170.09 49.96 29.95 3.00 3.00 35.05 16.41 3758.91
28 0 0 0 0 170.09 49.96 29.95 3.00 3.00 34.85 16.48 4352.41
29 0 0 0 0 170.09 49.96 29.95 3.00 3.00 35.00 16.17 3897.65
30 0 0 0 0 170.09 49.96 29.95 3.00 3.00 35.15 16.32 4125.62
31 0 0 0 0 170.09 49.96 29.95 3.00 3.00 35.18 16.21 4126.56

Y1, sensory score; Y2, solids loss (%); Y3, texture.

 2004 Blackwell Publishing Ltd International Journal of Food Science and Technology 2004, 39, 191–200
196 Optimization of pasta made from sweet potato S. Singh et al.

The mapping of the fitted responses was

Analysis of data
achieved using Stat-Ease software (Stat-Ease,
Multiple regression analysis was used to fit the version 6.01, 2001, Minneapolis, MN, USA).
model, represented by an equation, to the experi- The response surfaces and contour plots for these
mental data. Maximization and minimization of models were plotted as a function of two variables,
the polynomial thus fitted was done by numerical while keeping the other two variables at the
techniques, using the mathematical optimizer optimum level. The overlapping of the topographs
procedure of Quattro Pro software package was done to take into account the two responses
(Quatro Pro 1989, version 5.0; Borland Interna- for their optimum values corresponding to two
tional, Inc., Scotts Valley, CA, USA) that deals variables at a time.
with constraints. The constraints are set to get the
coded value of a variable between the lower and
Results and discussion
upper limits (in the present case: )2 and +2) for
an optimum response (a minimum and a maxi-
Diagnostic checking of the fitted models
mum level must be provided for each parameter
included). A weight can be assigned to each goal All main effects, linear and quadratic, and
to adjust the shape of its particular desirability interaction were calculated for each model. The
function. The goals are combined into an overall regression coefficients are shown in Table 4, as
desirability function. The program seeks to max- well as the correlation coefficient obtained for
imize this function. The goal seeking begins at a each model. The correlation coefficients for the
random starting point and proceeds up the steep- responses (R2 ¼ 0.928, 0.947 and 0.847, respect-
est slope to a maximum. There may be two or ively) are quite high for response surfaces and
more maximums because of curvature in the indicated that the fitted quadratic models accoun-
response surfaces and their combination into the ted for more than 85% of the variance in the
desirability function. By starting from several experimental data, which were found to be highly
points in the design space chances improve for significant. Based on t-statistics, the only regres-
finding the ÔbestÕ local maximum. sion coefficients significant at 95 and 99% levels

Table 4 Estimated coefficients of

Estimated coefficients the fitted quadratic equation for
different responses based on
Y1 Y2 Y3
t-statistic
Factors Coefficients t-Value Coefficients t-Value Coefficients t-Value

X0 35.033** 186.063 16.422** 37.643 4048.220** 9.280

X1 0.020 0.197 0.065 0.277 )374.823 )1.591
X2 )0.819** )8.056 )0.843** )3.580 85.908 0.365
X3 )0.653** )6.425 1.443** 6.125 1208.731** 5.131
X4 )0.310** )3.049 )0.526* )2.231 )560.643* )2.380
X12 0.190 2.039 1.054** 4.881 87.762 0.407
X22 0.064 0.683 0.257 1.193 )77.417 )0.359
X32 )0.249* )2.671 1.520** 7.043 746.116** 3.457
X42 )0.353** )3.785 1.011** 4.686 134.303 0.622
X12 0.416** 3.342 )1.643** )5.695 )711.299* )2.465
X13 0.315* 2.529 2.131** 7.385 1445.253** 5.009
X14 0.104 0.833 )0.266 )0.922 )447.850 )1.552
X23 )0.755** )6.062 )1.713** )5.935 675.281* 2.340
X24 )0.206 )1.656 )1.507** )5.224 403.076 1.397
X34 0.250 2.007 0.280 0.971 209.105 0.725
R2 0.928 0.947 0.847

*P < 0.05 (2.145), **P < 0.01 (2.977) for d.f. ¼ 14.
The t-values in bold indicate the significance at different P levels.

International Journal of Food Science and Technology 2004, 39, 191–200  2004 Blackwell Publishing Ltd
 Optimization of pasta made from sweet potato S. Singh et al. 197

were selected for developing the models as Table 5 Analysis of variance for different models
follows:
Sources of Sum of Mean
Sensory score ðY1 Þ ¼ 35:03  0:82X2  0:65X3 Responses variation d.f. squares square F

 0:31X4  0:25X23  0:35X24 þ 0:42X1 X2 Y1 Regression 14 50.929 3.638 14.659**

þ 0:32X1 X3  0:75X2 X3 Residual 16 3.971 0.248
Total 30 54.900
d.f. ¼ 14; R2 ¼ 0:928; Y2 Regression 14 381.545 27.253 20.457**
Solids loss ðY2 Þ ¼ 16:42  0:84X2 þ 1:44X3 Residual 16 21.315 1.332
Total 30 402.860
 0:53X4 þ 1:05X21 þ 1:52X23 þ 1:01X24 Y3 Regression 14 1.18E + 08 8.45E + 06 6.34**
 1:64X1 X2 þ 2:13X1 X3 ¼ 1:71X2 X3 Residual 16 2.13E + 07 1.33E + 06
Total 30 1.40E + 08
 1:51X2 X4 d.f. ¼ 14; R2 ¼ 0:947;
Texture ðY3 Þ ¼ 4:48:22 þ 1208:73X3 *P < 0.05 (2.37), **P < 0.01 (3.45) for d.f. ¼ 14.

 560:64X4 þ 746:12X23
 711:30X1 X2
þ 1445:25X1 X3 þ 675:28X2 X3 Conditions for optimum responses
d.f. ¼ 14; R2 ¼ 0:847:
The models (Y1, Y2 and Y3) were useful for
The sweet potato flour ratio had a greater positive indicating the direction in which to change vari-
quadratic effect on solids loss. The water ratio had ables in order to maximize sensory score and
a negative linear effect on both the responses. The texture and minimize solids loss. The multiple
soyflour ratio had a negative linear and quadratic regression equations were solved for the maxi-
effect on the sensory score, but a positive linear and mum sensory score (36.60), texture (9687 g) and
quadratic effect on solids loss. The Arabic gum minimum solids loss (13.34%). The optimum
ratio had a negative linear and quadratic effect on ingredientsÕ ratios (coded) predicted for each
sensory score and a negative linear and positive corresponding response are shown in Table 6,
quadratic effect on solids loss. The interaction of and lay within the experimental range, indicating
sweet potato flour and water content ratio had a the validity of the selection of the variables range.
significant positive and negative effect on sensory The calculated actual compositions (g kg)1) were:
score and solids loss, respectively. The interaction 656.79 sweet potato flour, 188.63 water, 127.41
of sweet potato flour and soyflour ratio had a soyflour, 12.81 Arabic gum and 14.36 CMC for
positive effect on both the responses. The interac- sensory score, 675.32 sweet potato flour, 195.32
tion of water content and soyflour ratio had a water, 105.88 soyflour, 12.66 Arabic gum and
negative effect on both the responses. The interac- 10.82 CMC for solids loss, and 666.11 sweet
tion of water content and Arabic gum ratio had a potato flour, 194.59 water, 119.16 soyflour, 9.22
negative effect on solids loss because of greater Arabic gum and 10.92 CMC for texture.
absorption and binding. The response surfaces and contour graphs were
obtained by plotting two variables, with the two re-
Analysis of variance maining having the values which give the optimum

When a model had been selected, an analysis of

variance was calculated to assess how well the Table 6 Predicted coded levels of ingredients yielding
maximum sensory score (36.60), texture (9687.0 g) and
model represented the data. An analysis of vari-
minimum solids loss (13.34%)
ance for all the responses is presented in Table 5.
The F-values for sensory score, solids loss and Levels for optimum response
texture (14.659, 20.457 and 6.341, respectively)
Ingredient’s ratio Sensory score Solids loss Texture
were significant at the 99% level. On this basis, it
can be concluded that the selected models X1 )0.663 0.988 0.150
adequately represent the data for sensory score, X2 )1.697 1.149 0.070
X3 )0.620 )0.981 1.834
solids loss and texture. There were no outliers to
X4 )0.541 0.852 )0.779
the regressions.

 2004 Blackwell Publishing Ltd International Journal of Food Science and Technology 2004, 39, 191–200
198 Optimization of pasta made from sweet potato S. Singh et al.

20.50 20.30

Solids loss (%)

Solids loss (%)
19.25 19.04
18.00 17.79
16.53
16.75
15.27
15.50

16.11
116.11
16.11

19.46
2.00 16.85
16.95 2.00

17.79
2.00 19.46 2.00 1.00 1.00
1.00 19.46 1.00 0.00 0.00
17.79
0.00 0.00 Water ratio –1.00 –1.00
Arabic gum ratio –1.00 -1.00 Water ratio –2.00 –2.00 Sweet potato flour ratio
–2.00 –2.00
Plate A
Plate A

20.50

Solids loss (%)

19.25
18.00
37.35 16.75
Sensory score

36.06 15.50
34.77
33.48 16.11
32.19
36.08
2.00 16.95 19.46 2.00
1.00 1.00
32.19 34.77
0.00 0.00
–2.00
2.00 Soy flour ratio –1.00 –1.00 Water ratio
–1.00 –2.00 –2.00
1.00
0.00 0.00
Plate B
Water ratio –1.00 1.00 Soy flour ratio
–2.00 2.00 Figure 3 Effect of ingredients on solids loss of pasta
product.
Plate B
Figure 2 Effect of ingredients on solids loss (plate A) and The response surface of solids loss affected by
sensory score (plate B) of pasta product. sweet potato flour and water ratio (Fig. 3, plate A)
and soyflour and water ratio (Fig. 3, plate B)
exhibited an Ôasymmetrical saddleÕ; contours were
response in the equations of Y1, Y2 and Y3. Some asymmetrical hyperbolas. There was a ÔminimaxÕ,
selected surfaces are presented in Figs 2–4. i.e. a minimum at the bottom of the saddle,
The relationship of solids loss with water and whereas the surface tended to rise at the edges
Arabic gum ratio is shown in Fig. 2 (plate A). The (Edger & Himmelblau, 1988). The same effect was
unit change in Arabic gum ratio does not produce observed by Rustom et al. (1991) in the extraction
much change in response, whereas, when moving of peanut protein.
in the direction of water ratio, a change in ratio A region of minimum hardness was observed in
brings greater changes in solids loss. The response Fig. 4 (plate A) and the partial derivatives showed
surface for sensory score was saddle-shaped, and that the independent variables (coded) for this
similar values were encountered at various com- minimum were 0.150 sweet potato flour ratio and
binations of minimum and maximum levels of the 1.834 soyflour ratio. These conditions are almost
independent variables (Fig. 2, plate B). near those for pasta, which resulted in maximum

International Journal of Food Science and Technology 2004, 39, 191–200  2004 Blackwell Publishing Ltd
 Optimization of pasta made from sweet potato S. Singh et al. 199

2.00

7000.00
1.00
Hardness (g)

6250.00

Water ratio
5500.00 33.83
(Sensory score)
4750.00 0.00
4000.00 16.57%
Solids loss

–1.00
5616.70 g
Hardness
2.00 3413.86
4499.99 2.00 –2.00
1.00
4862.03 1.00 –2.00 –1.00 0.00 1.00 2.00
0.00 Plate A
0.00
Soy flour ratio –1.00 Sweet potato flour ratio
–1.00
Sweet potato flour ratio
–2.00 –2.00
Plate A
Plate A
2.00

33.83
(Sensory score)
1.00
7000.00

Arabic gum ratio

Hardness (g)

5616.70 g
6250.00 Hardness
5500.00
0.00
4750.00
4000.00
–1.00
16.57 %
Solids loss
2.00 3413.86
3775.90 2.00 –2.00
1.00 –2.00 –1.00 0.00 1.00 2.00
4137.95 1.00
0.00 0.00 Soy flour ratio
–1.00 4862.03 –1.00
Soy flour ratio Water ratio
–2.00 –2.00 Plate B

Plate B Figure 5 Superimposed contour plots for all the responses

affected by the ingredients level.
Figure 4 Effect of ingredients on texture (hardness) of pasta
product.
were selected so that all three responses met their
observed response. This suggested that the two optimum acceptable region with the same ingre-
variables could be related. dient levels. It was suggested by the panel that
The dependence of hardness on variables (equa- sensory score (32–36.50), texture (4000–7000 g)
tion Y3) may be observed in Fig. 4 (plate B). There and solids loss (13.57–18.92%) were acceptable
is a slight decrease then a strong increase in within this range. Superimposing the individual
the hardness as soyflour ratio increases, slightly contour plots for the response variables resulted in
reduced by the negative and small interaction with the identification of a region (shown by the
water ratio. hatched area) that satisfied all constraints. At this
superimposition, the predicted values of sensory
score, solids loss and texture were 33.83, 16.57%
Superimposition of contour plots of responses
and 5616.7 g force, respectively. For these predic-
Areas of optimum performance were located by ted responses, the coded values of variables were
superimposing contour graphs of all the three first decoded into actual ratios as per the equa-
responses, using ingredient levels that established tions in Table 3 and then these ratios (x1,…,x4)
limits of acceptable quality for each factor, as were transformed into actual variables (A–E ) by
shown in Fig. 5 (plates A and B). As the optimum solving the algebraic equations as described in the
processing variables for each response did not fall section ÔExperimental designÕ. The final predicted
in exactly the same region in the two-dimensional optimum conditions for all the three responses are
space formed by the ingredient levels, constraints shown in Table 7.

 2004 Blackwell Publishing Ltd International Journal of Food Science and Technology 2004, 39, 191–200
200 Optimization of pasta made from sweet potato S. Singh et al.

Table 7 Optimized values for sensory score (33.83), texture Colonna, P., Doublier, J.L., Melcion, J.P., Monredon, F.D.
(5616.7 g) and solids loss (16.57%) after superimposition & Mercier, C. (1984). Extrusion cooking and drum drying
of contour curves for different variables of wheat starch. I. Physical and macromolecular mod-
ifications. Cereal Chemistry, 61, 538–543.
Ingredients (g kg)1) Edger, T.F. & Himmelblau, D.M. (1988). Formulation of
objective functions. In: Optimization of Chemical Pro-
A. Sweet potato flour 673.95
cesses (edited by B.J. Clark & J.M. Morriss). Pp. 78–81.
B. Water 195.10
New York: McGraw-Hill.
C. Soyflour 110.28
El-sahy, K.N. & Siliha, H. (1988). Use of sulphite treated
D. Arabic gum 10.59
sweet potato flour in manufacture of wheat bread.
E. CMC 10.08
Getreide Mehl und Brot, 42, 215–217.
Total 1000.00 Hamed, M.G.E., Siliha, H. & Sandy, S.K. (1973). Prepara-
tion and chemical composition of sweet potato flour.
Cereal and Bakery Products, 50, 133–139.
The optimum conditions were experimentally Iwe, M.O. & Onuh, J.O. (1992). Functional and sensory
tested, obtaining a sensory score of 35.4, a texture property of soybean and sweet potato flour mixture.
Lebensmittel Wissenschaft und Technologie, 25, 569–573.
of 5589 g force and a solids loss of 17.1%. These
Jauregui, R.N.C., Silva, M.E.M.P. & Areas, J.A.G. (2000).
experimental values at the optimum compositions Extrusion cooking process for amaranth (Amaranthus
were in good agreement with the predicted values. caudatus L.). Journal of Food Science, 65, 1009–1015.
Limroongreungrat, K. & Huang, Y.W. (2002). Develop-
ment of pasta products using alkaline-treated sweet
Conclusions potato flour fortified with soy protein. In: Abstract
Volume: 2002 Annual Meeting and Food Expo, Anaheim,
RSM was successfully applied to the determin- California, 16–19 June. p. 46H-5 (http://ift.confex.com/
ation of the optimum formulation of pasta ift/2002/techprogram/meeting_2002.htm).
prepared using sweet potato flour as a main in- Lizada, M.L.C. & Guzman, M.P. (1982). Processing of flour
gredient. Products made from sweet potato flour from sweet potato. UP Home Economics Journal, 10, 62–
73.
by extrusion may be functional alternatives to
Ministry of Food Processing Industries (2003). Annual
wheat-based products for consumers who cannot Report. p. 24. New Delhi: Department of Food Process-
tolerate wheat gluten. However, consumer testing ing Industries, Ministry of Food Processing Industries.
is recommended to further validate the accepta- Onwulata, C.I., Konstance, R.P., Strange, E.D., Smith,
bility of the sweet potato for pasta products. P.W. & Holsinger, V.H. (2000). High-fiber snacks
extruded from triticale and wheat formulations. Cereal
Foods World, 45, 470–473.
References Rosell, C.M., Rojas, J.A. & De Barber, C.B. (2001).
Influence of hydrocolloids on dough rheology and bread
Bouwkamp, J.E. (1985). Sweet Potato Products: a Natural quality. Food Hydrocolloids, 15, 75–81.
Resource for Tropics. Pp. 139–147. Bacaraton, FL: CRS Rustom, I.Y.S., Lopez-Leiva, M.H. & Nair, B.M. (1991).
Press, Inc. Optimization of extraction of peanut proteins with water
Centre Proposals and TAC Recommendations (2000). by response surface methodology. Journal of Food
CGIAR Research – Area of Research. Washington, DC: Science, 56, 1660–1663.
Consultative Group on International Agricultural Sebio, L. & Chang, Y.K. (2000). Effects of selected process
Research (http://www.cgiar.org/index.html). parameters in extrusion of yam flour (Dioscorea rotunda-
Chiappe, L., Lopez, P.E. & Villagarcia, M. (1984). Yield ta) on physicochemical properties of the extrudates.
and flour production of twelve white sweet potato Nahrung-Food, 44, 96–101.
cultivars for industrial use. In: Proceedings of Sixth Thakur, S. & Saxena D.C. (2000). Formulation of extruded
Symposium of the International Society for Tropical Root snack food (gum based cereal-pulse blend): optimization
Crops (edited by J.E. Wanham). Pp. 587–589. Lima, Peru: of ingredients levels using response surface methodology.
Faculty of Agriculture, National Agrarian University. Lebensmittel Wissenschaft und Technologie, 33, 354–361.
Cocharan, W.G. & Cox, G.M. (1957). Some methods for Vatsala, C.N., Saxena, D.C. & Rao, P.H. (2001). Optimiza-
the study of response surface. In: Experimental Designs tion of ingredients and process conditions for the
(edited by R.A. Bradley, D.G. Kendall, J.S. Hunter & preparation of puri using response surface methodology.
G.S. Watson). Pp. 335–375. New York: John Wiley and International Journal of Food Science and Technology, 36,
Sons. 407–414.

International Journal of Food Science and Technology 2004, 39, 191–200  2004 Blackwell Publishing Ltd