b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Available online at www.sciencedirect.com

ScienceDirect
journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Effects of sunflower meal quality on the technical

parameters of the pelleting process and pellet
quality


c a,*, Lato L. Pezo b, uro M. Vukmirovic a,
Radmilo R. Colovi


c a, Oskar J. Bera c, Vojislav V. Banjac a, Jovanka D. Levic a
Du
sica S. Colovi
a

Institute of Food Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia
Institute of General and Physical Chemistry, University of Belgrade, Studentski Trg 12 e 16, 11000 Belgrade, Serbia
c
Faculty of Technology, University of Novi Sad, Bulevar cara Lazara 1, 21000 Novi Sad, Serbia
b

article info

The impact of sunflower meal quality (i.e. crude protein content and crude fibre content) on

Article history:

the technical parameters of the pelleting process and the physical properties of produced

Received 2 July 2015

pellets was investigated. Five model mixtures were prepared for pelleting, with different ratios

Received in revised form

of corn, sunflower meal (SFM) and soybean meal (SBM). Three grades of sunflower meal were

25 September 2015

used in the experiments with crude protein contents of approximately 37%, 40%, and 43%.

Accepted 30 September 2015

Within each of the mixtures, granulation of material and retention time in the steam condi-

Published online 11 November 2015

tioner were varied. In order to describe the effects of the test variables on the observed responses response surface methodology, standard score analysis and principal component

Keywords:

analysis (PCA) were used. The increase in the protein content and the decrease in the crude

Sunflower meal

fibre content of sunflower meals caused an increase in the pelleting temperature, specific

Pelleting

energy consumption, pellet hardness, and pellet durability. In addition, an increase in

Physical quality

retention time increased specific energy consumption of pellet press, and produced fines. In

Technological parameters

terms of pellet durability values, a longer retention time was more beneficial for SFM mixtures
than for SBM mixture. The type of mixture was found to be the most influential variable for
second order polynomial model calculation. Standard score analysis showed that the optimum values for energy consumption, quantity of the fines, and pellet durability indices were
obtained for the mixture with 40% protein SFM, with no retention after conditioning and with
the finest granulation of the components (0.933). PCA showed that the first two principal
components (91.10% of the total variability) enabled a neat separation of the five mixtures.
2015 IAgrE. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

The composition of pelleted feed is driven by nutritive demands of the animals. However, the choice of ingredients is to

some extent flexible. Thus, it is necessary to know the technical impact of selected raw materials on pellet quality. If a
compound mixture with technically unacceptable ingredients
is pelleted, the pellets obtained may be of poor quality
(Buchanan & Moritz, 2009; Thomas, Rijm, & Van der Poel,

* Corresponding author. Tel.: 381 (0)21 485 37 96; fax: 381 (0)21 45 07 25.


).
E-mail address: radmilo.colovic@fins.uns.ac.rs (R.R. Colovi
c
http://dx.doi.org/10.1016/j.biosystemseng.2015.09.010
1537-5110/ 2015 IAgrE. Published by Elsevier Ltd. All rights reserved.

99

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Nomenclature
Symbols
d
dHMSO
dgw
E
E0
Esp
Fs
H
PDI
Q
Rt
Sgw
Td

Pellet diameter (mm)

Diameter of hammer mill screen openings (mm)
Geometric mean diameter (mm)
Energy consumption during pelleting of
experimental mixtures (kW)
Energy consumption of pellet press with no
material added (kW)
Specific energy consumption (kW/t/h)
Proportion of fines (%)
Pellet hardness (N)
Pellet durability index
Pellet press feeder rate (kg/h)
Retention time (s)
Geometric standard deviation (mm)
Pellet press die temperature ( C)

2000). Cereals, which are rich in starch, are mostly used to

provide the required energy for domestic animals and they
represent the largest proportion of compound mixtures.

1.1.

Protein meals in pelleting diets

Numerous references describe the positive impact of protein

on pellet quality (Abdollahi, Ravindran, and Svihus, 2013;
Maier & Briggs, 2000; Winowiski, 1988). Leguminous and
oilseed meals are commonly used as a source of protein in
compound mixtures for monogastric animals, with soybean
meal (SBM) being most frequently used. Hence, the effect of
the addition of SBM on physical pellet quality was described in
several papers (Briggs, Maier, Watkins, & Behnke, 1999; Moritz
et al., 2002). Sunflower meal (SFM) is a feedstuff that can be
used as both as a source of protein and amino acids for
monogastric animals (Rehman, Asad, Qureshi, & Iqbal, 2013).
The high concentration of methionine, and its low concentration of lysine, allow SFM to be combined with SBM enabling
the formulation of diets without the addition of synthetic
amino acids (de Oliveira et al., 2014). Depending on the technical process used for oil extraction from sunflower kernel,
and additional technologies for protein enrichment, the protein and crude fibre content of SFM can vary in wide range of
concentrations (Briones, Serrano, & Labidi, 2012; Mushtaq
et al., 2009). Although a widely used ingredient, little
research on the physical quality of SFM pellets is available.
Israelsen, Busk, and Jensen (1981), as well as MacMahon and
Payne (1991), estimated the effects of different raw materials
on physical quality of SFM pellets. According to their estimates and classification reports, SFM can be regarded as
having a more positive effect on pellet physical quality than
SBM. However, within these estimates there was no data on
the applied process parameters.

1.2.

Objectives of the present study

Although there are a few reports comparing SFM to other raw

materials in terms of its impact on the physical quality of

Tp
X
Y

Pellet temperature ( C)
Variable
Response

Greek symbols
b
Constant regression coefficient
Statistical abbreviations
ANOVA Analysis of Variance
PCA
Principal component analysis
Coefficient of determination
r2
RSM
Response surface methodology
SOP
Second order polynomial
SD
Standard deviation
SS
Standard score
Abbreviations
SBM
Soybean meal
SFM
Sunflower meal

pelleted feed, differences in SFM quality have not yet been

taken into consideration. Therefore, the main objective of this
study was to investigate the impact of SFM quality (i.e. crude
protein and crude fibre content) on the pellet quality and the
dependent technical parameters of the pelleting process.
Additionally, the influence of the independent process variables, the diameter of the hammer mill screen openings (dHMSO)
and retention time in the steam conditioner (Rt), on the physical
quality of the pellets and the technical parameters of the process was examined for each of the model mixtures examined.

2.

Materials and methods

2.1.

Raw materials and model mixtures

Raw materials used in this study were corn, SFM and SBM.
Corn, sort NS 300, was grown in the northern province of
Vojvodina, Serbia. SFM was provided by local oil producing


factory Victoria Oil, Sid,
Serbia, while SBM was provided by

ej, Serbia. The chemisoy processing factory Sojaprotein, Bec
cal composition of oilseed meals is shown in Table 1. Three
grades of SFM were provided, differing in crude protein and
crude fibre content: SFM 1, SFM 2 and SFM 3 with the crude
protein content of approx. 37%, 40%, and 43% respectively.
The listed materials were used to prepare five model mixtures. The model mixtures were used so a comparison of the
impact of specific oilseed meals on the pellet quality and the

Table 1 e Chemical composition of oilseed meals used in

the experiment (dry matter basis).
Composition (g kg1)
Crude protein
Crude fibre
Crude ash
Crude fat
Nitrogen free extract

SFM 1

SFM 2

SFM

SBM

374
207
80
27
312

403
192
86
28
291

427
178
83
37
275

472
27
62
13
426

100

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

parameters of the pelleting process could be made whilst

minimising the effects of other ingredients. The composition
of the model mixtures is presented in Table 2. Mixture C
consisted only of corn, mixtures SFM 37, SFM 40, and SFM 43
consisted of corn and specific sunflower meal, while mixture
SBM 47 contained corn and soybean meal with a 47% of protein. The corn-to-oilseed meal ratio in mixtures SFM 37, SFM
40, SFM 43, and SBM 47 was 6.5:3.5, which corresponds to the
proximate mean ratio of cereals and oilseed meals in compound mixtures for broilers and turkeys (Sauvant, Perez, &
Tran, 2004). The experimental mixtures were formulated to
investigate the impact of specific oil seeds on the technical
parameters of the pelleting process and pellet quality.

stock of sieves on a laboratory sieving device (Endecotts, UK).

Geometric mean diameter (dgw) and geometric standard deviation (Sgw) were determined to evaluate particle size and
particle size distribution according to ASAE (2006) standard.
Pellet press die temperature was measured by a Pt 100
temperature probe placed to touch outer die wall, while the
temperature of the hot pellets collected at exit of the press
was measured using a contact thermometer (model DM
9231A, Transfer Multisort Elektronik, Poland).
Specific energy consumption (Esp kW t1 h1) during pelleting was calculated according to the Eq. (1):

2.2.

where E (kW) is energy consumption during pelleting of

experimental mixtures, E0 is the energy consumption of pellet
press with no material added, and Q (kg h1) is pellet press
feeder rate.
The proportion of fines (%) in pelleted mixtures was
determined by sieving approximately 100 g of pelleted material through a 4.8 mm screen.
The diameter of pellets (d) was measured by using micrometer from a random sample of 15.
Hardness of the pellets was determined by using texture
analyser (model TA.HD Plus, Stable Micro Systems Ltd, UK) by
measuring the first peak force (N) during diametric compression. For hardness determination, 15 pellets of approximately
equal length (12 mm) from each of pelleted mixtures were
collected. Load cell of 500 N and 45 mm cylindrical probe were
used for diametric compression.
A Pfost pellet tester (Buhler, Switzerland) was used for
pellet durability measurement, expressed as a pellet durability index (PDI). In each of rotating boxes of the device 500 g
of pellets was placed. Rotating speed was set at 50 rpm, and
duration of rotation was set at 10 min. After the rotation,
pellets were removed from the boxes and sieved through
4.8 mm screen. The ratio of mass of pellets after the test to
pellets before the test was expressed as PDI.

Processing

Processing of the materials was performed at the Feed to Food

pilot plant, Institute of Food Technology, Novi Sad, Serbia. The
corn was ground in a laboratory hammer mill (Model 11, ABC
Engineering, Serbia). The hammer mill was equipped with
three screens with 2, 3, and 4 mm diameter openings. Protein
meals were pre-ground in the same laboratory hammer mill
but no screens were used in order to crush agglomerates. A
twin-shaft pedal mixer/steam conditioner (SLHSJ0.2, Muyang,
China) was used for mixing and steam conditioning/ripening
of the material. Mixing time for the dry mixture was set at 60 s,
after which 200 kPa pressure saturated steam was injected into
the dry mixture until material reached the temperature of
80  C. The Rt in the mixer/steam conditioner was varied as
follows: preheated material was discharged from the mixer
upon reaching the temperature of 80  C (zero retention time) or
it was retained at maximum temperature for 5 and 10 min. The
moisture level of the material after steam conditioning/
ripening was set to be ~16%. A flat die laboratory pellet press
(14-175, Amandus Kahl, Germany), with drive capacity of 3 kW
and two rolls, was used to pellet model mixtures. The parameters of the pellet press were: die thickness 24 mm, diameter of
openings 6 mm (length/diameter ratio 4:1), distance rolls-die
0.5 mm, roll speed 135 rpm, and throughput 18 kg h1. The
cooling of the hot pellets to a room temperature was performed
in a vibrating cooler (FB 500  200, Amandus Kahl, Germany).

2.3.

Chemical and physical analyses

Dry matter content, crude protein, crude fibre, crude fat, crude
ash, and starch content in the model mixtures were determined according to AOAC official methods (AOAC, 2000).
Sieving analysis was performed according to ISO 1591-1
(1988) standard by sieving approx. 100 g of material, using

Table 2 e Formulation of the model mixtures.

C
Ingredients (g kg1)
Corn
1000
SFM 1
e
SFM 2
e
SFM 3
e
SBM
e

SFM 37

SFM 40

SFM 43

SBM 47

650
350
e
e
e

650
e
350
e
e

650
e
e
350
e

650
e
e
e
350

Esp

2.4.

E  E0
$1000 ;
Q

(1)

Experimental design

The experimental data used for the study of experimental

results were obtained using a 5  32 full factorial experimental
design (5 mixtures and within each 2 parameters at 3 levels),


, Cvetkovic
, & Kovac

evic
,
according to RSM (Pezo, Suput,
Levic
2014). Independent experimental factors for each of the five
mixtures are shown in Table 3.

2.5.

Statistical analysis

Descriptive statistical analyses of all the obtained results have

been expressed using mean and standard deviation (SD) for
each of the model mixtures. Collected data have been subjected to ANOVA to explore the effects of process variables.
Furthermore, principal component analysis (PCA) was applied
to classify and discriminate the different mixtures. A pattern
recognition technique was applied within result descriptors to
characterise and differentiate all observed samples. The
evaluation of response surface methodology (RSM), ANOVA

101

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Table 3 e Independent experimental factors and their

levels.
Experimental factor

Symbol

Diameter of hammer mill

screen openings (mm)
Retention time in the
steam conditioner (min)

Table 4 e Chemical composition of the experimental

mixtures (dry matter basis).

Coded factor levels

1
(Low)

0
(Centre)

1
(High)

X1

X2

10

Composition (g kg )
Crude protein
78.7
Crude fibre
34.5
Crude ash
14.9
Crude fat
41.8
Nitrogen free extract 830.1
713.4
Starcha
a

and PCA analyses were performed using StatSoft Statistica

10.0 software (Statistica, 2010).
A numbers of PCA biplots tests were defined using the
following equation (Eq. (2)):
N1

Rt
3$dHMSO  2 9$MN  1
5

(2)

where Rt is 0, 5 or 10; dHMSO is 2, 3 or 4; MN is 1 for corn, 2 for

SFM 37, 3 for SFM 40, 4 for SFM 43 and 5 for SBM 47.
The following second order polynomial (SOP) model was
fitted to the experimental data. Seven models of the following
form were developed to relate seven responses (Y) and two
process variables (X), for each of five different model mixtures
(Eq. (3)):
Ykl blk0

2
X

blki $Xi

i1

2
X

blkii $X2i blk12 $X1 $X2 ; k 1  7; l 1  p;

i1

(3)
blk0 ,

blki ,

blkii ,

blk12

where:
are constant regression coefficients; Ykl ,
either: pellet press die temperature Td, pellet temperature Tp,
specific energy consumption Esp, proportion of fines Fs, pellet
diameter d, pellet hardness H, and pellet durability index PDI
with X1 diameter of hammer mill screen openings dHMSO and
X2 retention time in the steam conditioner Rt.
In order to have a more complete overview of the ranking
of observed model mixtures, SS were evaluated using a chemometric approach by experimentally measured responses.
Minimumemaximum normalisation is one of the most
widely used technique to compare various characteristics of
complex samples determined by using multiple measurements,
where samples are ranked based on the ratio of raw data and the
extreme values of the measurement used (Brlek et al., 2013).
Since the units and the scale of the data from various physical
and chemical characteristics differ, the data in each data set
were transformed into normalised scores. The sum of normalised scores of a sample of different measurements when averaged give a single non-dimensional value (SS) which is a specific
combination of data from different measuring methods with no
limitation on units. This approach could help comparisons with
others set of data obtained from future measurements.

3.

Results and discussion

3.1.

Chemical composition and particle size

The chemical composition of the experimental mixtures is

shown in Table 4.

SFM 37 SFM 40 SFM 43 SBM 47

1

181.0
94.9
38.8
33.3
652.0
464.9

190.8
89.6
39.9
33.5
646.2
467.7

200.6
84.6
39.1
36.8
638.9
458.8

214.2
31.8
30.1
31.7
692.2
445.9

Consisted in Nitrogen free extract.

The results in Table 4 show that the mixture comprising

only of corn (C) had, as expected, the highest content of starch,
and the lowest contents of crude protein and crude ash when
compared to other mixtures. Mixtures SFM 37, SFM 40, and
SFM 43 had similar chemical compositions. Since SFM was
included in these three mixtures, the crude protein content
and crude fibre content were higher than mixture C. Moreover, the crude protein content increased and crude fibre
content decreased from mixture SFM 37 to mixture SFM 43,
which is in line with the chemical composition of SFM used in
a specific mixture. Mixture SBM 47 had the highest protein
content of all five mixtures since SBM had a higher protein
content than SFM. This mixture had considerably lower crude
fibre and crude fat content.
The geometrical mean diameter (dgw) and geometric standard deviation (Sgw) of the experimental mixtures are shown
in Table 5. Mixtures with oilseed meals had smaller dgw than
mixture C, since they contained finely ground oilseed meal.
Slight differences in granulation between mixtures SFM 37,
SFM 40, and SFM 43 were a result of SFM quality. SFM 1 had
high share of coarse particles originating from hulls, while
SFM 3 had the highest proportion of protein rich fine particles
from the kernels, due to separation of hulls by using screens in
the industrial production of high protein SFM. There were no
significant differences (p > 0.05) in Sgw for the different mixtures and within the same dHMSO used for the milling of corn.
However, significant differences (p < 0.05) between Sgw values
were observed within the specific mixture and for different
dHMSO. In theory, minimal geometrical standard deviation is
equal to 1, which is the case if all particles of the material have
the same size (ASAE, 2006). Values of Sgw showed that particles
were more equally distributed in size when a smaller screen
size was used.

3.2.
Comparison of treatment means, descriptive
statistics and PCA
The descriptive statistics of the pellet press production parameters and pellet physical quality parameters for each of
the mixtures tested are shown in Table 6.
The values of Td as well as Tp depended on the mixture
composition (Table 6). The highest Td and Tp were obtained for
the mixture C (Td 67.53  C and Tp 62.49  C), while the
lowest temperatures were obtained for the mixture SFM 37
(Td 57.53  C and Tp 56.73  C). Examining the results for the
SFM mixtures, it is obvious that the increase in the crude

102

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Table 5 e Granulation parameters of the experimental mixtures.

Granulation parameter

Diameter of screen openings (mm)

SFM 37

SFM 40

SFM 43

SBM 47

2
3
4
2
3
4

625.26
839.98
925.20
1.819
2.019
2.190

497.73
619.01
659.80
1.880
2.169
2.350

487.76
604.37
644.44
1.867
2.116
2.251

476.46
588.17
628.65
1.882
2.257
2.283

620.95
734.75
769.05
1.874
2.038
2.123

dgw (mm)

Sgw (mm)

dgw e geometric mean diameter, Sgw e geometric standard deviation.

protein content and the decrease in crude fibre content caused

an increase in Td and Tp. Temperature differences between the
experimental mixtures can be explained by differences in the
water absorption capacities of the mixtures. Hemmingsen
et al. (2008) showed that the water absorption of cereal com
ponents was higher than water absorption of SBM, while Levic
 (2012) showed that water absorption of SBM
and Sredanovic
was higher than water absorption of SFM. High water absorption possibly caused less water available at the surface of
the material. The decrease of water on the surface of the
particles increases friction in the pellet press channels, and
increases pellet press die temperature (Moritz et al., 2001).
Thus, the highest temperature was observed for the pelleting
mixture C, while Td and Tp were lower for the pelleting of SFM
mixtures. The decrease of Td and Tp with an increase of crude
fibre content within SFM mixtures was also a result of more
water on the surface of the particles (Briggs et al., 1999).
Jonsson, Kuri, Tudorica, and Brennan (2003) and Zimonja,
, Edvardsen, and Svihus (2008) show that
Hetland, Lazarevic
insoluble fibres absorb water very slowly, which is one of the
reasons for more water was available on the surface of particles with more crude fibre content in SFM mixtures, and
consequently a temperature decrease.
All pelleted samples (45 samples in total) were produced
from different combination of the experimental mixtures, and
various Rt and dHMSO values, as shown by experimental design,
and predicted by the PCA score plot (Fig. 1). The PCA allowed
for a considerable reduction in a number of variables required
and the detection of structure in the relationship between the

measuring parameters. The full auto scaled data matrix consisting of five different mixtures with different technical
treatment was submitted to the PCA.
A scatter plot of samples using the first two principal
components (PCs) issued from PCA of the data matrix was
obtained (Fig. 1) in order to visualise the data trends and the
discrimination efficiency of the descriptors used. Quality results showed that the first two principal components, accounting for 91.10% of the total variability can be considered
sufficient for data representation.
As can be seen in Fig. 1, there was a distinct separation
between the five mixtures, according to assays used. The
general impact of the independent experimental factors (Rt
and dHMSO) can be also observed. Retention of the mixtures in
the mixer/steam conditioner affected Td and Tp which
increased with increasing of Rt. This can be seen in Fig. 1,
where Rt, Td and Tp are positioned on the same side of the PCA
biplot. The reason for the temperature increase with longer Rt
was the better penetration of water into the particles and the
consequent increase of friction in the pellet press die channels. Briggs et al. (1999) increased conditioning time from 5 to
15 s by changing the angle of the steam conditioner pedals.
These authors stated that increase of steam conditioner
retention time caused the temperature to rise across the die
due to decrease of material surface moisture.
The results from Table 6 show that the highest energy
consumption was observed for the mixture C, while the
lowest specific energy consumption was observed for the
mixture SFM 37, which is also in line with the differences in

Table 6 e Descriptive statistics of the production parameters and pellet physical quality parameters (n 9).
Mixt. type
Ave.
SD
Ave.
SD
Ave.
SD
Ave.
SD
Ave.
SD
Polarity

C
SFM 37
SFM 40
SFM 43
SBM 47

Td

Tp

Esp

Fs

PDI

67.53
2.94
57.53
1.46
58.11
2.02
60.13
1.92
66.18
1.76

62.49
1.70
56.73
0.96
56.91
1.53
58.32
1.79
62.17
1.90

30.63
2.12
15.58
1.62
15.43
1.76
18.14
2.17
28.44
2.56
e

2.75
0.44
1.89
0.62
1.29
0.37
1.19
0.22
1.78
0.47
e

5.98
0.04
6.06
0.03
6.04
0.03
5.98
0.03
5.79
0.03

119.54
12.26
113.27
14.51
136.90
12.85
161.12
13.83
231.63
28.24

87.72
0.60
89.39
2.29
93.56
2.05
95.26
0.80
95.62
0.63

Polarity: the higher the better criteria,  the lower the better criteria (n 9).
Ave.average, SDstandard deviation.

Td e pellet press die temperature ( C), Tp e pellet temperature ( C), Esp e specific energy consumption during pelleting of experimental mixtures
(kW), Fs e share of fines (%), d e pellet diameter (mm), H e pellet hardness (N), PDI e pellet durability index.

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Fig. 1 e PCA biplot for pelleting parameters and pellet

quality. Td e pellet press die temperature ( C), Tp e pellet
temperature ( C), Esp e specific energy consumption during
pelleting of experimental mixtures (kW), Fs e proportion of
fines (%), d e pellet diameter (mm), H e pellet hardness (N),
PDI e pellet durability index, dHMSO e diameter of hammer
mill screen openings (mm), Rt e retention time (min), SS e
standard score (/).

friction in the die channels for pelleting of different mixtures.

For the pelleting of the SFM mixtures, decrease in insoluble
fibre content caused an increase of the specific energy
consumption.
The increase in the Rt caused an increase in the specific
energy consumption (Fig. 1). This is also a consequence of the
increase in the friction in the die channels.
The average values of pellet hardness for all five mixtures
were 119.5, 113.3, 136.9, 161.1, and 231.6 N, respectively (Table
6). Generally, pellet hardness increased with the increase in
the protein content of the mixture. The exception was the
mixture C which had higher hardness than the mixture SFM
37. Although mixture C had lower protein content than
mixture SFM 37, it also had lower crude fibre content (34.5
compared to 94.9 g kg1). Insoluble fibres (such as fibres from
sunflower hull) are resilient, and thus pellets can have weak
spots, which may result in their fragmentation (Kaliyan &
Morey, 2009).
Increase of Rt caused a decrease in pellet hardness (Fig. 1).
The reason may be the decrease of surface water (already
shown by Td, Tp and Esp differences), which negatively affected
liquid necking of the neighbouring particles (Aarseth &
Prestlkken, 2003; Rumpf, 1958). Coarser milling caused
lower hardness of the pellets (opposite direction of Factor 2
coordinate from the PCA biplot), also due to the appearance of
weak spots, i.e. spots where most probable breakage of the
pellet can occur (Thomas, Van Zuilichem, & Van der Poel,
1997).
PDI values (Table 6) increased with the increase of protein
content of the mixtures. Although mixture C had a lower
crude fibre content than mixture SFM 37, it also had a higher

103

content of starch and lower content of crude protein, causing

lower PDI values of mixture C when compared to mixture SFM
37. Zimonja and Svihus (2009) added native starch to the diet
for broilers and also showed that the addition of starch
negatively influenced pellet quality, and that higher PDI
values of non-starch diet can be attributed to the protein
content. Additionally, Calvacanti (2004) indicated that the
corn proteins have a negative impact on pellet quality. As
expected, an increase in the protein content and a decrease in
crude fibre content within SFM mixtures led to increase of PDI
values, which corresponds to the results of Wood (1987) and
Winowiski (1988), who had shown that native proteins in the
pelleting process act as a binder. Moderate temperatures may
result in destruction of the three-dimensional structure of
proteins, and they may result in new covalent-bond formations (Svihus & Zimonja, 2011). Additionally, PCA biplot (Fig. 1)
has shown that PDI for the mixtures was highly correlated
with protein content and inversely correlated with fat and
native starch content.
Figure 1 shows that Rt differently affected the PDI of the
mixtures C and SBM 47, when compared to the mixtures with
SFM (mixtures SFM 37, SFM 40, and SFM 43). For the mixtures C
and SBM 47 it was observed that PDI values decreased with
increase in Rt, probably due to the decrease in surface moisture. On the other hand, for the SFM mixtures the maximum
PDI values were observed for the Rt of 5 min. It appears that an
Rt of 5 min was optimal in terms of sufficient chemical alterations in material and sufficient surface moisture. Generally,
mixture SBM 47 had the highest PDI values of all mixtures for
Rt of 0 min (conventional steam conditioning without retention), while the mixtures SFM 40 and SFM 43 (with higher
quality SFM) had higher PDI values for 5 and 10 min retention
time.
MacMahon and Payne (1991) classified different raw materials according to their physical and chemical characteristics, as well to their response in the pelleting process. These
authors rated sunflower meal (33.9% of protein (as fed))
contribution to the pellet quality with grade 6 (out of 10), while
soybean meal (44% of protein (as fed)) was rated with grade 4,
which is in contrary to the results of this study comparing
similar raw materials (mixture SFM 37 and mixture SBM 47).
Namely, mixture SBM 47 had higher PDI values when
compared to mixture SFM 37 for all values of dHMSO and Rt.
However, it should be recalled that pellet quality also depends
on other factors of feed mill, as already described by Reimer
(1992). It is to expect that values of the feed mill factors from
the present study differed from the values from which
MacMahon and Payne (1991) obtained their results.
As expected, increasing dHMSO negatively affected pellet
durability (Fig. 1). Coarser particles have a lower specific surface which negatively affects the binding between them, as
already described by other authors (Franke & Rey, 2006).

3.3.

Analysis of variance and standard score analysis

An analysis of variance (ANOVA) was conducted to show the

significant effects of independent variables on the responses,
and to show which responses were significantly affected by
the varying treatment combinations (Table 7).

104

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Table 7 e ANOVA table.

Sum of squares

dHMSO
d2HMSO
Rt
Rt2
Mixture
Mixture2
dHMSO  Rt
dHMSO  Mixture
Rt  Mixture
r2

Td

Tp

Esp

Fs

PDI

99.07
7.44
32.92
2.55
0.01
723.23
4.90
3.22
3.31
0.899

32.53
5.33
8.31
5.55
0.82
268.89
12.39
9.51
9.26
0.883

1.04
0.67
159.51
0.71
2.96
1844.38
0.41
1.52
2.11
0.946

0.45
0.00
5.27
0.08
6.20
7.38
0.46
0.00
0.04
0.903

7.86
5.62
1.32
32.97
422.36
16.51
2.22
0.07
0.50
0.875

535.35
301.36
128.57
7.94
6788.26
1560.24
43.44
13.44
43.05
0.966

0.00
0.01
0.00
0.02
0.18
0.21
0.00
0.00
0.00
0.965

Bold numbers denote values significant at p < 0.01 level, 95% confidence limit.
Td e pellet press die temperature ( C), Tp e pellet temperature ( C), Esp e specific energy consumption during pelleting of experimental mixtures
(kW), Fs e share of fines (%), d e pellet diameter (mm), H e pellet hardness (N), PDI e pellet durability index, dHMSO e diameter of hammer mill
screen openings (mm), Rt e retention time in the steam conditioner (min).

The SOP models for all variables were found to be statistically significant and the response surfaces were fitted to
these models. The most influential variable for SOP model
calculation was the type of mixture. The quadratic term of this
variable was most influential for Td, Esp, Fs and PDI calculation
(p < 0.01 level). The linear term of mixture type contributed the
most in the d and H calculation (p < 0.01 level). Linear terms of
dHMSO and Rt were very important in the Td, Tp, Fs and H
calculation (p < 0.01), while quadratic terms of dHMSO and Rt
have been very influential in PDI calculation (p < 0.01), while Rt
was influential in the Esp calculation (p < 0.01).
All SOP models (Table 7) had an insignificant lack of fit
tests, which means that all the models represented the data
in a satisfactory manner (Montgomery, 1984). A high r2 is an
indication that the variation was accounted and that the
data fitted satisfactorily to the proposed model (Madamba,
2002). The r2 values for Esp (0.946), Fs (0.903), H (0.966) and
d (0.965) were high, while r2 values for Td (0.899), Tp (0.883)
and PDI (0.875) could be considered as being satisfactory and
showed a good fit between the model and the experimental
results.
Standard score analysis (SS) was used in this experiment to
develop strategies for the improvement of final product
characteristics with lower production cost. SS is dimensionless value and is in consistent agreement with observed assays. Values of SS above 0.700 showed the high standard in the
desired process parameters and pellet properties (low Esp, low
Fs, and high PDI). Variations in dHMSO and Rt were introduced to
obtain the optimal values for pelleting process responses.
The SS analysis showed that the optimum values for Esp, Fs,
and PDI characteristics were obtained for mixture SFM 40, for
Rt of 0 min and dHMSO of 2 mm (0.933); this can be seen from
Fig. 1. Samples located on the left of the PCA biplot graphic
showed better SS results (optimisation according to Fs, Esp, and
PDI), which were in the range 0.80e0.94. Generally, the SFM 40
mixture had the highest number of high SS results (above
0.645, average was 0.798), followed by mixture SFM 43 (average
was 0.764), while mixtures C, SFM 37 and SBM 47 had the
generally low SS results (averages were 0.199, 0.674 and 0.478,
respectively).

4.

Conclusion

The results of this study showed that the difference in crude

protein and crude fibre content of SFM caused differences in
technical parameters of pellet press as well as in the pellet
physical quality. The increase in crude protein content slightly
increased the pelleting temperature and specific energy consumption but caused an improvement of pellet durability and
pellet hardness. Inclusion of high protein SFM (43%) in the
experimental diets allowed production of pellets with same
physical quality as with SBM, but with lower production costs.
The longer retention time in the steam conditioner was more
beneficial for diets with SFM than for diets with SBM in terms
of pellet durability.

Acknowledgement
This article is produced within the project III 46012, funded by
the Serbian Ministry of Education, Science and Technological
Development.

references

A.O.A.C. (2000). Official methods of analysis of AOAC International.

Arlington, VA.
Aarseth, K. A., & Prestlkken, E. (2003). Mechanical properties of feed
pellets: weibull analysis. Biosystems Engineering, 84(3), 349e361.
Abdollahi, M. R., Ravindran, V., & Svihus, B. (2013). Pelleting of
broiler diets: an overview with emphasis on pellet quality and
nutritional value. Animal Feed Science and Technology, 179(1),
1e23.
ASAE standards 319.3. (2006). Method of determining and expressing
fineness of feed materials by sieving. ASABE standards, 608. St.
Joseph, MI: American Society of Agricultural and Biological
Engineers.
Briggs, J. L., Maier, D. E., Watkins, B. A., & Behnke, K. C. (1999).
Effect of ingredients and processing parameters on pellet
quality. Poultry Science, 78(10), 1464e1471.

b i o s y s t e m s e n g i n e e r i n g 1 4 0 ( 2 0 1 5 ) 9 8 e1 0 5

Briones, R., Serrano, L., & Labidi, J. (2012). Valorization of some

lignocellulosic agro-industrial residues to obtain biopolyols.
Journal of Chemical Technology and Biotechnology, 87(2), 244e249.


a, N., Kric

ka, T., Vukmirovic
, ., Colovi
, R.,
Brlek, T., Pezo, L., Voc
c
et al. (2013). Chemometric approach for assessing the quality
of olive cake pellets. Fuel Processing Technology, 116, 250e256.
Buchanan, N. P., & Moritz, J. S. (2009). Main effects and
interactions of varying formulation protein, fiber, and
moisture on feed manufacture and pellet quality. The Journal of
Applied Poultry Research, 18(2), 274e283.
Cavalcanti, W. B. (2004). The effect of ingredient composition on the
physical quality of pelleted feeds: A mixture experimental approach.
Doctoral dissertation. Kansas State University.
Franke, M., & Rey, A. (2006). Buhler, AG Uzwill, Switzerland,
Improving pellet quality and efficiency. Feed Tech, 10(3).
Hemmingsen, A. K. T., Stevik, A. M., Claussen, I. C.,
Lundblad, K. K., Prestlkken, E., Srensen, M., et al. (2008).
Water adsorption in feed ingredients for animal pellets at
different temperatures, particle size, and ingredient
combinations. Drying Technology, 26, 738e748.
ISO 2591-1. (1988). Test sieving-part 1: Methods using test sieves of
woven wire cloth and perforated metal plate. Geneva:
International Standards Organization.
Israelsen, M., Busk, J., & Jensen, J. (1981). Pelleting properties of
dairy compounds with molasses, alkali-treated straw and
other byproducts. Feedstuffs, 7, 26e28.
Jonsson, J. M., Kuri, V., Tudorica, C. M., & Brennan, C. S. (2003).
Dietary fibres: their uses in dairy based applications. In Dietary
Fibre: Bio-active carbohydrates for food and feed. Dietary Fibre 2003
Conference. The Netherlands: Wageningen Academic
Publishers, Wageningen University.
Kaliyan, N., & Morey, R. V. (2009). Factors affecting strength and
durability of densified biomass products. Biomass and
Bioenergy, 33(3), 337e359.
, J., & Sredanovic
, S. (2012). Sunflower meal e monograph. Novi
Levic
Sad: Institute of Food Technology.
MacMahon, M. J., & Payne, J. D. (1991). The pelleting handbook.
Sarpsborg Norway: Borregaard Lignotech.
Madamba, P. S. (2002). The response surface methodology: an
application to optimize dehydration operations of selected
agricultural crops. LWT-Food Science and Technology, 35(7),
584e592.
Maier, D. E., & Briggs, J. L. (2000). Making better. Feed and Grain, 1,
12e15.
Montgomery, D. C. (1984). Design and analysis of experiments (2nd
ed.). New York: John Wiley and Sons Inc.
Moritz, J. S., Beyer, R. S., Wilson, K. J., Cramer, K. R.,
McKinney, L. Y., & Fairchild, F. J. (2001). Effect of moisture
addition at the mixer to a corn-soybean based diet on broiler
performance. Journal of Applied Poultry Research, 10, 347e353.
Moritz, J. S., Wilson, K. J., Cramer, K. R., Beyer, R. S.,
McKinney, L. J., Cavalcanti, W. B., et al. (2002). Effect of
formulation density, moisture, and surfactant on feed
manufacturing, pellet quality, and broiler performance. The
Journal of Applied Poultry Research, 11(2), 155e163.

105

Mushtaq, T., Sarwar, M., Ahmad, G., Mirza, M. A., Ahmad, U.,
Noreen, U., et al. (2009). Influence of sunflower meal based
diets supplemented with exogenous enzyme and digestible
lysine on performance, digestibility and carcas responce of
broiler chickens. Animal Feed Science and Technology, 149(3e4),
275e286.
de Oliveira, T. M. M., Nunes, R. V., Marcato, S. M., Berwanger, E.,
ne, R. A., et al. (2014). Sunflower meal for
Bayerle, D. F., Scho
broilers: chemical composition and metodologics
amendments on determination of energetic values and
rias (Londrina),
digestible aminoacids. Semina: Ci^encias Agra
35(6), 3415e3427.


, Lj B., Cvetkovic
, B. R., &
Pezo, L. L., Suput,
D. Z., Levic

evic
, O. A. (2014). Effects of temperature and immersion
Kovac
time on rehydration of osmotically treated pork meat. Journal
of Food and Nutrition Research, 53(3), 260e270.
Rehman, T., Asad, F., Qureshi, N. A., & Iqbal, S. (2013). Effect of
plant feed ingredients (soybean and sunflower meal) on the
growth and body composition of Labeo rohita. American Journal
of Life Sciences, 1(3), 125e129.
Reimer, L. (1992). Conditioning. In Proc. Northern crops institute feed
mill management and feed manufacturing technology e Short
course, 7, CPM Crawfordsville, USA.
Rumpf, H. (1958). Grundlagen und methoden des granulierens.
Chemie Ingenieur Technik, 30(3), 144e158.
Sauvant, D., Perez, J. M., & Tran, G. (2004). Tables of composition and
nutritive value of feed materials: Pigs, poultry, cattle, sheep, goats,
rabbits, horses, fish. Paris: Versailles: Wageningen Academic
Publishers, Wageningen and INRA Editions.
STATISTICA (Data Analysis Software System). (2010). v. 10., StatSoft, Inc., USA, (www.statsoft.com).
Svihus, B., & Zimonja, O. (2011). Chemical alterations with
nutritional consequences due to pelleting animal feeds: a
review. Animal Production Science, 51(7), 590e596.
Thomas, M., Rijm, W., & Van der Poel, A. F. B. (2000). Functionality
of raw materials and feed composition. In Proceedings feed
manufacturing in the Mediterranean Region, Reus, Spain (pp. 1e24).
Thomas, M., Van Zuilichem, D. J., & Van der Poel, A. F. B. (1997).
Physical quality of pelleted animal feed. 2. Contribution of
processes and its conditions. Animal Feed Science and
Technology, 64(2), 173e192.
Winowiski, T. (1988). Wheat and pellet quality. Feed Management,
39(9), 58e64.
Wood, J. F. (1987). The functional properties of feed raw materials
and their effect on the production and quality of feed pellets.
Animal Feed Science and Technology, 18(1), 1e17.
, N., Edvardsen, D. H., &
Zimonja, O., Hetland, H., Lazarevic
Svihus, B. (2008). Effects of fibre content in pelleted wheat and
oat diets on technical pellet quality and nutritional value for
broiler chickens. Canadian Journal of Animal Science, 88(4),
613e622.
Zimonja, O., & Svihus, B. (2009). Effects of processing of wheat or
oats starch on physical pellet quality and nutritional value for
broilers. Animal Feed Science and Technology, 149(3), 287e297.